AI has the same problem, at ten times the clock speed.

By the end of this read, you will have a working financial model that maps your full token cost stack, isolates hidden consumption, and tells you whether your pricing survives a provider rate shock before the market does.

The free deep dive told you what to model. This one hands you the model.

If you have not read “The New Math of AI,” start there. It covers the four paths, the token iceberg, and why the fastest-growing AI startups are running at 25% gross margins while the healthier cohort operates closer to 60%. This post assumes you understand the problem. It is here to solve it.

What follows is a walk-through of the Token Waterfall Model: a six-sheet working financial model built specifically for AI operators who need to price a product before launch, not discover their cost structure after. Every number in this article maps directly to a cell in the model. Open both. Work through them together.

What This Model Does That Generic SaaS Models Do Not

Every financial model template on the market was built for SaaS. Revenue minus COGS, COGS being hosting and support, gross margin being 80%. That math does not apply here.

The Token Waterfall Model is built around four realities that are specific to AI products:

Token costs are not fixed. They vary by user behavior, task complexity, model version, and provider pricing decisions. A model that assumes a flat cost-per-user will be wrong by 2x to 10x depending on your product.

Hidden token consumption is the primary cost driver. Internal reasoning chains, agent loops, retry logic, and tool calls often represent 50-90% of total token spend. Almost no standard model accounts for this.

Provider rate changes are a strategic risk, not a line item. Pricing tier restructuring by major AI providers has in some documented cases resulted in materially higher infrastructure costs for enterprise clients within a single contract period. Your model needs a shock test built in before you need it.

Power users destroy average-based pricing. The math that works for your median user can be catastrophic when applied to your top 5%. You need to know which segments are profitable and which are subsidized before you set a single price.

The Token Waterfall Model addresses all four. Here is how each sheet works.

Sheet 1: The Token Flow Calculator Where every pricing model should start. Almost none do.

Before you can price anything, you need to know what a single transaction actually costs in tokens. Not what the user prompt costs. Not what the response costs. Everything.

The Token Flow Calculator maps your product’s full token consumption across two layers.

Read more

I have spent nearly two decades building commercial systems where pricing and cost structure were not finance problems, they were survival problems. At Aurora, I ran strategic growth and partnerships in a market where mismodeling the cost of a financing product by even a few basis points could destroy a deal. That same pattern is playing out right now in AI, at a much faster clock speed.

By the end of this read, you will understand what your AI product actually costs to deliver, why that number is probably two to five times what you think, and how to build the financial model that closes the gap.

Every company building with AI right now is on one of four paths.

Which one you are on determines your cost structure, your margin profile, and your strategic exposure. Most companies are on path one. Most have no idea what path one actually costs them.

The fastest-growing AI startups are running at 25% gross margins. Sometimes negative. The healthier cohort operates closer to 60%. Traditional SaaS sits at 80-90%. Bessemer Venture Partners documented this spread in their State of AI 2025 report [1]. The cause is physics: every inference call has a price. Every generated sentence incurs real compute cost that traditional SaaS never faced.

Here is what each path actually means.

1. Path one: use an existing API. Fast, flexible, and someone else’s infrastructure decision. Also someone else’s pricing decision, their rate hike, and increasingly their competing product built on the same stack. One important note: path one on frontier models carries rising cost exposure. GPT-4o and Claude 3.5 class models cost more per token than their predecessors in many use cases, particularly for long-context and reasoning-heavy tasks. The inference cost curve is falling for older models. It is not falling uniformly across the stack.

2. Path two: run open source models on your own infrastructure. Rent the compute, own the weights. Lower margin risk than API dependence, lower capital than a proprietary build. The fastest-growing path among sophisticated builders right now, for exactly the margin reasons described in this piece.

3. Path three: customize a foundation model on your own data. More control, more cost, and a longer runway to production.

4. Path four: build proprietary from scratch. A serious 30-billion parameter model starts at $5-10 million in compute alone. A frontier-class model costs north of $100 million to train. Billion-dollar runs are coming before the end of 2026.

Most companies are on path one. Most have not yet modeled what path one actually costs them per transaction, fully loaded, including every token their product consumes that a user never sees.

That is where the margin goes.

The Token Is the New Unit of Production

In traditional SaaS, the marginal cost of serving one more user approaches zero. In AI, it does not.

A token is roughly 0.75 words or 4 characters. A typical user interaction costs 200-300 tokens on the surface. The real number is often ten times that once you account for the full system. This is the Token Iceberg, and most operators only price for the visible tip.

Before you set a single price, answer one question: what is your actual token flow? That means every system prompt, every reasoning step, every agent loop and retry your product runs in the background. In agentic AI products, internal token consumption is 50-90% of total usage, the cost operators most often fail to model.

The breakdown matters more than the total. A product with 800 tokens of visible consumption and 7,200 tokens of internal consumption does not have a usage problem. It has a pricing problem, because the cost structure is almost certainly invisible to the team setting the price.

Make sure the Math is Math-ing.

Work through your own numbers with the Token Flow Calculator at kylekelly.co/tools/token-calculator.

What Surviving Companies Did Differently

Three pricing models have emerged. Each represents a different trade-off between vendor cost predictability and customer value alignment.

As you move from consumption to outcome-based pricing, you accept more cost risk in exchange for tighter alignment with customer value. Most early-stage AI companies default to consumption because it is the easiest to implement. It is also the easiest for customers to abandon when spend becomes unpredictable.

The most defensible commercial position is outcome-based pricing, but only if you have fully loaded your cost model first. Intercom charges $0.99 per resolved support ticket. That model works because they built the cost structure before setting the price, not after. Build the model wrong and outcome pricing becomes a liability, not an advantage.

Subscribe now

The Credit Is an Abstraction Layer, Not a Cop-Out

Credits exist because the true unit of AI production (the token) is too granular for customers to reason about. A credit is a pricing wrapper that converts technical consumption into commercial clarity. But credits solve a communication problem, not a financial one. Your internal cost structure still runs on tokens.

Here is how three representative products have structured it:

The Manus Standard tier reveals the gap most clearly. At 0.5 cents per credit, with a cost-to-serve closer to 0.85 cents, they are selling below cost at current pricing. That is a deliberate land-and-expand play. It only works if two things happen before the runway runs out: engagement deepens enough to justify a price increase, and volume grows enough to negotiate better infrastructure rates. Most companies running this playbook will not hit both.

Cursor’s margin profile is contested. Multiple credible sources reported negative gross margins in mid-2025, with significant API costs against subscription revenue. Their trajectory is a useful signal regardless of the exact number: even products with strong PMF and rapid growth can face structural margin pressure when token consumption scales faster than pricing catches up.

The Four Numbers That Determine Whether You Make It

A rigorous operator does not set price from intuition. They build a model that converts token costs into commercial decisions. There are four layers to that model, and most operators only think about one.

Layer 1: Revenue Architecture

The revenue line is more complex than it looks. Subscription revenue anchors too low in almost every product I have analyzed, because operators underestimate how much credit consumption scales with engagement. The operators getting this right are building four revenue streams, not one: subscription, overage, tier upsells, and enterprise contracts. Each has a different margin profile. Each needs to be modeled separately before you set a blended price.

Layer 2: Cost of Goods Sold

Three line items most operators miss entirely.

Inference costs include every token type your product consumes, not just the visible input and output. Internal reasoning steps, agent loops, and retry logic are all billed at the same rate as user-facing tokens. If you have not instrumented this, you do not know your COGS.

Infrastructure overhead runs 10-15% of inference cost annually across a typical AI product stack: hosting, monitoring, logging, security. This is not a rounding error at scale.

Third-party API costs are the line item teams most often forget. Every search call, vector database query, or external tool call in your agentic workflow has a cost. Even $0.002 per transaction compounds significantly at 200,000 monthly transactions.

Your gross margin is the gap between what you charge per credit and what every token in that credit actually costs you to produce. That gap is almost always smaller than operators expect.

Layer 3: The Gross Margin Reality

The 60% gross margin threshold matters because it is the level at which LTV:CAC math becomes defensible for most institutional investors. It is a target, not a gate. The Bessemer Supernova cohort runs at 25% and still attracts capital because growth rate can temporarily substitute for margin in early stages. [2] The question is not whether you are at 60% today. It is whether you have a credible path to it before your runway forces the conversation.

If you are selling credits at $0.005 and your cost-to-serve is $0.0085, your gross margin is negative before you pay for a single sales call or engineer. Model the full waterfall before you set a price.

Better pricing buys you runway. It does not fix a broken cost structure at the model layer. That requires volume, routing, or your own model. But most operators are not close to that problem yet. They are still losing margin to costs they have not measured.

Layer 4: The Questions That Expose Your Real Risk

Most pricing conversations focus on what to charge. These four questions focus on what you are not accounting for. Operators who answer them before launch live in a different risk category than those who discover them after.

What is the fully loaded cost of your most expensive task type, including abnormal reasoning chains? Most cost models are built on average usage. Power users and edge cases destroy average-based pricing.

Do you currently lose money on your heaviest users, and do you know who they are? If you do not have this number, you do not have a cost model. You have a guess.

What happens to your gross margin if your model provider raises API costs 20%? This is not a hypothetical. Anthropic introduced Priority Service Tiers in 2025 that doubled infrastructure costs for some enterprise clients overnight. If your pricing does not survive a 20% cost shock, you have a dependency, not a strategy. This is also where open source infrastructure changes the math permanently.

What external APIs or agents are embedded in your standard workflow that you are not yet billing against? Every third-party tool call in your product stack has a cost. Most operators only discover this at scale.

Where This Is Going

The companies sustaining 60% gross margins share one behavior: they modeled the full cost stack before pricing, not after. The operators who survive will not do so by building better models. They will do so by building better pricing systems.

The token is the ingredient. The outcome is the product. Price accordingly.

References

1. State of AI, Bessemer Venture Partners, Sep 2025

2. The Gross Margin Debate in AI, Tanay Jaipuria, Sep 2025

3. Cursor Pricing, 2025

4. Manus Pricing, 2025

5. Perplexity Pricing, 2025

6. GPT-4 Training Cost, Sam Altman / Stanford AI Index, 2024

7. Frontier Model Cost Projections, Dario Amodei / Fortune, Apr 2024

8. Inference and Compute Costs, Scale AI Foundation Model Report, Aug 2025

9. Custom LLM Build Costs 2025, Follow.life / Epoch AI Analysis, 2025

For the last two decades, software has been eating the world, but it has mostly been consuming the same meal: the knowledge economy. The physical economy (the industries that build, move, and power our world) has been left with the scraps. This massive half of global GDP, representing over $50 trillion in annual output, remains a frontier of operational complexity, held together by fragmented point solutions and manual processes. Motive Technologies’ recent S-1 filing for its IPO provides a detailed blueprint for how to build a category-defining company by systematically digitizing these foundational industries. It’s a case study in strategy, systems thinking, and the power of a full-stack approach to AI.

I wrote previously about why compliance is becoming one of the most durable moats in modern software. Motive is a live case study of that theory playing out at scale. This is a compliance moat that compounds across regulation, insurance economics, and data defensibility.

A System of Record for the Physical World

Motive’s core premise is that the physical economy is not just a large market, but a fundamentally different one that requires a new kind of platform. Unlike the natively connected knowledge economy, the physical world runs on disconnected assets: vehicles, equipment, and mobile workers. This creates immense challenges, from safety and labor shortages to rising costs and regulatory pressure. Motive’s strategy is to build the Integrated Operations Platform, a unified system of record that connects and automates a customer’s entire physical footprint (workers, vehicles, equipment, and spend).

The platform architecture is deceptively simple: six interconnected modules (Driver Safety, Fleet Management, Equipment Monitoring, Spend Management, Workforce Management, and AI Vision) that share a common data foundation. This integration is the key differentiator. Rather than forcing customers to stitch together point solutions from multiple vendors, Motive provides a single pane of glass for physical operations.

Data from the S-1 Filing

Motive’s S-1 is a testament to the scale of the opportunity and the effectiveness of their strategy. The company is not just growing; it is accelerating its hold on the most valuable customer segments. The filing shows a clear pattern of moving upmarket and increasing customer dependency on the platform.

As of September 30, 2025, Motive serves nearly 100,000 customers. The growth in high-value cohorts is particularly telling. Large Customers, those with an Annual Recurring Revenue (ARR) over $100,000, grew by 58% year-over-year, and their contribution to total ARR jumped from 28% to 37%. This demonstrates a successful land-and-expand model that deepens wallet share over time.

This growth is built on solving acute, expensive problems. The physical economy is grappling with systemic risks that directly impact profitability. For instance, fatal crashes involving large trucks surged 43% from 2014 to 2023, and commercial trucking insurance premiums have risen approximately 40% over the past decade. Simultaneously, organizations face a complex and ever-changing regulatory landscape. Navigating this environment is not just an operational burden; it is a strategic imperative where building an ELD compliance moat can become a significant competitive advantage. Motive’s platform directly addresses these issues, turning compliance from a cost center into a streamlined, data-driven function.

The AI-Powered Flywheel

Motive’s competitive moat is not just its software, but its full-stack, AI-powered system. The company has engineered a powerful, recursive flywheel that creates compounding value and raises barriers to entry.

This system is built on four pillars:

1. Proprietary Hardware: Devices like the AI Dashcam and Vehicle Gateway are purpose-built to capture high-fidelity, multimodal data (video, telematics, diagnostics) in harsh operating environments.

2. Massive, Labeled Data Corpus: Over one million connected vehicles and assets feed a continuously growing data set. A team of 400 full-time annotators labels tens of millions of events annually, creating a rich, structured foundation for model training.

3. Purpose-Built AI Models: This data is used to train highly accurate, domain-specific AI models. A 2023 study by the Virginia Tech Transportation Institute found Motive’s AI Dashcam was 2-4x more effective at detecting unsafe driving behaviors than its competitors.

4. Low-Latency Validation: A tight feedback loop allows for rapid validation of model outputs, eliminating false positives and ensuring the high degree of accuracy required for mission-critical operations.

The risk in this model is not competition, but execution drag. Hardware margins compress, annotation costs scale non-linearly, and regulatory regimes evolve faster than product cycles. A full-stack moat only compounds if the company continues to translate model accuracy into auditable, insurer-grade outcomes. Lose that translation layer and the flywheel degrades into an expensive data operation. This is why compliance is not a feature, but the control plane.

This flywheel (where more customers lead to more data, which leads to better AI models, which in turn attracts more customers) is the engine of Motive’s defensibility. It’s a classic data network effect applied to the physical world.

How to Build for the Physical Economy

If your AI product can be removed without breaking a compliance workflow, you are not building a system of record.

Motive’s S-1 offers a clear playbook for founders and executives aiming to build enduring companies in industrial and vertical AI markets.

1. Target a Fragmented, High-Stakes Workflow: Identify a core operational process that is currently managed through a patchwork of disconnected point solutions. For Motive, this was fleet management and safety compliance.

2. Build the System of Record First: Before chasing sophisticated AI features, build the foundational data layer. Motive’s “Physical Operations Graph” unifies data from across a customer’s operations, creating a single source of truth and high switching costs.

3. Go Full-Stack to Control the Data Pipeline: In the physical world, data quality is paramount. Owning the hardware (the sensor layer) ensures the fidelity of the data feeding the AI models. This control over the end-to-end system is a powerful differentiator that pure software players cannot replicate.

4. Translate AI Accuracy into Economic Value: Don’t just sell AI; sell outcomes. Motive quantifies its impact in terms of accident reduction (80% average for AI Dashcam users), insurance savings (~25% for top quartile customers), and fuel and fraud savings ($175M+ in 2024). This makes the ROI tangible and the purchasing decision straightforward.

The Industrialization of AI

Motive’s IPO is more than a financial event; it’s a signal that the next wave of enterprise software will be built on the bedrock of the physical economy. The playbook is shifting from pure-play SaaS to integrated systems that combine hardware, software, and purpose-built AI. These are not just software companies, but systems companies. They are building the digital infrastructure that will automate and optimize the industries that form the backbone of our civilization.

Motive has provided the blueprint. Now, a generation of founders will execute on the same playbook.

I’m Kyle Kelly. Each week Line of Sight breaks down how AI, strategy, and revenue growth architecture turn complexity into leverage.

This is not investment advice.

References

1. Motive Technologies, Inc. (2025, December 23). Form S-1 Registration Statement. U.S. Securities and Exchange Commission.

There is a class of billion-dollar startups that do not compete on features. They compete on becoming the invisible, indispensable infrastructure for navigating regulatory complexity. I call this strategy The Compliance Moat.

A compliance moat is a defensible competitive advantage built by embedding a product so deeply into a regulated workflow that it becomes the de facto standard. The moat is not the product itself; it is the network of integrations, the accumulated data, and the high switching costs created by solving a complex, fragmented, and mandatory compliance problem.

Compliance moats are not permanent monopolies. They are time-bound leverage windows created by regulatory fragmentation, and they only become durable if converted into data ownership, distribution control, or workflow lock-in before standards stabilize.

This pattern is visible across four key industries. By examining them, we can uncover a repeatable playbook for building a compliance-driven business.

Four Verticals: Proving the Pattern

1. Crypto Tax Compliance: The Race to Become the System of Record

The average crypto user has assets across 10-20 different exchanges and wallets, creating a data aggregation nightmare for tax reporting [1]. This fragmentation fuels the crypto tax software market, projected to grow from $210.6 million in 2025 to $683.9 million by 2035 at a 12.5% CAGR [2].

Companies like CoinTracker, Taxbit, and Koinly are racing to become the single source of truth. CoinTracker’s partnership with Intuit TurboTax is a brilliant moat-building move, making it the default infrastructure for the largest tax software provider. As the IRS introduces Form 1099-DA for digital asset brokers in 2025, the company with the most comprehensive data network will become the undisputed system of record [3].

The USA, UK, Canada, Germany, Australia, France, and India already treat crypto like property or income (as of Dec 2025). This is a CAGR to keep an eye on as many countries clear up, evolve, or begin to enforce tax policies.

2. ADA Website Accessibility: Turning Legal Risk into a Sticky Product

The Americans with Disabilities Act (ADA) has created a litigious compliance landscape. With over 4,000 ADA-related lawsuits filed annually and a 37% year-over-year increase in the first half of 2025, businesses are desperate for a simple solution [4]. This legal pressure fuels a digital accessibility market projected to reach $3.24 billion by 2034 [5].

Companies like AudioEye and accessiBe build moats by embedding their tools into website builders like WordPress and Shopify. Once installed, the legal risk of removing the tool creates high switching costs, making the product a form of insurance against lawsuits.

The moat is not the regulation itself. It is the speed at which a company exploits regulatory complexity before it collapses into standardization.

3. IT Compliance Attestation: The Enterprise Sales Gatekeeper

In B2B SaaS, SOC 2 and ISO 27001 compliance are mandatory for selling to enterprise customers. This has created a cybersecurity certification market projected to grow from $3.88 billion in 2025 to $7.50 billion by 2030 [6].

Unicorns like Vanta ($4.15B valuation) and Drata ($2B valuation) have built moats by integrating with cloud providers like Amazon Web Services (AWS) and source code repositories like GitHub [7]. They become the system of record for a company’s security posture, making it incredibly painful to switch to a competitor for the next annual audit. The compliance certificate is the ticket to the enterprise sales game, and these platforms are the gatekeepers.

4. Logistics and Transportation: The Mandate-Driven Moat

The $14.5 trillion logistics industry is a prime example of a mandate-driven moat. The Electronic Logging Device (ELD) mandate requires commercial drivers to use electronic logs to track hours of service, creating an ELD market projected to grow from $15.1 billion in 2025 to $22.4 billion by 2035 [12].

Companies like Geotab and Samsara build moats by providing ELD solutions deeply integrated into trucking workflows. Switching costs are high, requiring hardware reinstallation and data migration. Similarly, freight brokers must maintain a $75,000 surety bond, fueling a transportation management system (TMS) market projected to reach $37.04 billion by 2030 [13] [14].

The Operator’s Dilemma: Build, Buy, or Partner?

I’ve made this decision from the inside, using this lens in live Build, Buy, and Partner decisions, including M&A transactions where compliance architecture materially affected deal outcomes.

Every new regulation forces a capital allocation decision. For example, when the EU AI Act lands, when crypto tax reporting becomes mandatory, when SOC 2 becomes table stakes for enterprise deals, operators face the same question: do we build this capability, buy a company that has solved it, or partner with infrastructure?

The decision hinges on two factors: whether compliance creates competitive differentiation, and whether a clear market leader has emerged.

Most companies partner. It is faster and cheaper than building, less risky than buying. But partnering creates dependency. Operators who partner without negotiating data rights, exit clauses, or migration paths trade short-term speed for long-term strategic exposure.

The exception: companies where compliance data itself becomes a competitive moat. Plaid owns bank connection infrastructure. Stripe owns payments compliance. For them, building was not optional. When compliance shapes product roadmap, pricing power, or platform trust, building is the only way to avoid ceding strategic control.

Buying is a high-risk, capital-intensive move reserved for strategic market consolidation or reclaiming control after a partnership proves too constraining.

AI: The Moat Builder & The Moat Breaker

The EU’s AI Act will create a $9.5B AI governance market by 2035, a 15.8% CAGR . Penalties up to 7% of global turnover will drive adoption. This creates a new generation of compliance moats around AI model auditing and validation.

But AI is also a moat breaker. It automates the complexity that sustains existing moats. AI can classify crypto transactions. It can audit websites for ADA compliance in real-time. It can monitor cloud infrastructure for SOC 2. Startups using AI will offer faster and cheaper solutions. The moats built on manual data aggregation are the most vulnerable.

Most incumbents will not be disrupted by better competitors. They will be disrupted when weeks of compliance work collapse into minutes and their accumulated friction evaporates.

The compliance moats that survive AI are not those built on process complexity, but those anchored in trust, audit authority, and irreversible workflow adoption.

The Question for the Next Decade

Compliance moats are not permanent. They require continuous reinvestment and adaptation. The tension is this: companies profit from complexity, but customers want simplicity. The best operators navigate this paradox. They make compliance easier for customers while building deeper moats. The question for the next decade of founders: are you building a new compliance moat, or are you the AI-powered startup that will tear the old one down?

For operators, the real advantage is not building better software. It is becoming the standard every other participant is forced to route through before regulation simplifies the game again.

Line of Sight delivers frameworks on AI, strategy, and growth systems for VP and C-suite operators. Written by Kyle Kelly. No theory. No hype. Just repeatable models that scale. Subscribe at lineofsight.io

References

1. CoinTracker. “2024 Crypto Tax Report”.

2. Future Market Insights. “Crypto Tax Software Market”.

3. Internal Revenue Service. “Digital Assets”.

4. EcomBack. “2025 Mid-Year ADA Website Lawsuit Report”.

5. Straits Research. “Digital Accessibility Software Market”.

6. Mordor Intelligence. “Cybersecurity Certification Market”.

7. Failory. “The Full List of 18 Compliance Unicorn Startups (2026)”.

8. European Commission. “Artificial Intelligence Act”.

9. European Commission. “Article 99: Penalties”.

10. Future Market Insights. “Enterprise AI Governance and Compliance Market”.

11. European Commission. “Article 16: Obligations of Providers of High-Risk AI Systems”.

12. Future Market Insights. “Electronic Logging Device Market”.

13. Federal Motor Carrier Safety Administration. “Broker and Freight Forwarder Financial Responsibility Rule Overview and Compliance”.

14. MarketsandMarkets. “Transportation Management System Market”.

I spent over a decade doing partnerships of every kind. Product partnerships are my favorite. They’re technical, challenging, very strategic, and when executed right, they drive customer acquisition, brand elevation, revenue expansion, and stickiness. They help a business run a faster race.

Product partnerships rarely fail in the announcement. They fail in measurement, ownership, and resourcing. We call this “value leakage,” but it’s simpler than that: it’s a failure of accountability. We treat partnerships like lottery tickets when we should treat them like systems.

This week, I’m breaking down where that value goes and what you can do about it. In my next post, I’ll share a 90-day playbook to reclaim it.

What you’ll get in the article

• How partnership value leaks before it ever hits the P&L (and the 5 failure modes to audit)

• A simple spectrum and matrix to pick the right partnerships, not just the exciting ones

The 5 guardrails to lock before engineering starts: economics, data rights, ownership, attribution, and exit

Product managers are measured on impact.

We ship product, drive metrics, and grow the business. But one of the most powerful levers for impact often remains underutilized or misunderstood: product partnerships. Too often, we approach partnerships with a value capture mindset, asking, “What can we get?” This is a mistake. The most successful partnerships start by asking, “What can we create, together?”

Marty Cagan of the Silicon Valley Product Group puts it best: “focus on value and always create more value than you capture.” [1] This is the fundamental principle that separates partnerships that merely extract value from those that create exponential growth.

Value creation is the strategy. Guardrails are the mechanism that prevents value creation from turning into free labor.

Create First, But Protect Capture

Focusing on value creation first does not mean ignoring the economics. It gives you permission to focus on building a bigger pie, but it doesn’t give you permission to accept weak terms. As an operator, you must establish guardrails to protect your ability to capture value later. Before any significant engineering work begins, ensure you have alignment on these five non-negotiable terms:

1. Minimum viable economics: “What is the smallest paid commitment on Day 1?”

2. Data rights: “Who can store, reuse, and market from shared data?”

3. Customer ownership: “Who supports, who renews, who can upsell?”

4. Attribution: “What is the source of truth and the event definition?”

5. Exit: “What happens to customers, data, and the integration on unwind?”

Securing these terms upfront prevents costly renegotiations and ensures that when you do create value, you have a clear path to capturing your fair share.

Start with Customer Jobs, Not Integration Features

Before diving into frameworks and playbooks, we need to establish the foundational principle: partnerships exist to solve customer problems, not to create technical integrations.

The most successful product partnerships work backwards from the customer. Identify complementary technologies working backwards from your customer and what they are trying to accomplish. Map the customer workflow and the jobs to be done, then identify where your product and your partner’s product can work together to eliminate friction.

This customer-first approach ensures you’re building integrations that solve real problems, not just technical exercises that look impressive on a partnership announcement but deliver no measurable value.

Where Partnership Value Can Miss the P&L

The reality is that most of the value created in a partnership is never realized. It leaks away. Let’s be clear: ‘friction’ and ‘misalignment’ are polite terms for unbudgeted costs that kill your ROI. The Integration Tax isn’t a metaphor; it’s a real, recurring expense that hits your engineering budget every quarter.(Note: the %s and timelines below are heuristics from my experience, used to help you size the problem.)

Based on my experience, I’ve developed a heuristic to visualize this: The Partnership Value Leakage Model. Imagine a partnership creates $100K in potential value. A significant portion of that value can be lost before it ever reaches the customer or the bottom line.

• Integration Tax: ongoing engineering build, maintenance, and debugging.

• Coordination Overhead: recurring meetings, roadmap syncs, and cross-team drag.

• Attribution Ambiguity: unclear source of truth, so both sides underinvest.

• Misaligned Incentives: priorities shift, commitments decay, execution stalls.

• Customer Friction: clunky handoffs and broken flows that drive churn.

As this illustrative chart shows, the majority of potential value can be lost to various forms of friction. The “Integration Tax” and “Misaligned Incentives” are often the biggest culprits. This is why a myopic focus on capturing a bigger slice of the pie is so dangerous. A bigger slice of a tiny, shrunken pie is still a tiny slice.

The Transactional Trap

The great trap for most organizations is getting stuck in transactional partnerships. They offer predictable, measurable, but linear returns. Leaders become addicted to the comfort of a simple rev-share model, and they lack the strategic patience to invest in the less-defined, but exponential, returns of deeper integrations. Most companies should earn the right to transformative partnerships through repeatable integrative wins.

Where to Focus: The Partnership Value Capture Matrix

Not all partnerships deserve equal investment. Use the Partnership Value Capture Matrix to decide where to focus your efforts. This 2x2 evaluates partnerships based on their value creation potential and the complexity of capturing that value.

Your goal is to spend most of your time on “Hidden Gems”, partnerships with high value creation potential and low capture complexity, like API and data-sharing agreements. “Ecosystem Plays” are your big bets, requiring significant investment but offering massive returns. Avoid “Strategic Traps” at all costs. These look impressive on paper but drain resources without delivering proportional value.

• Hidden Gems are high-value, low-complexity partnerships you can execute with your current resources.

• Ecosystem Plays require significant investment and infrastructure but unlock compounding, network-effect value.

• Strategic Traps aren’t bad partners, they’re bad timing. If you can’t capture the value yet, defer them until you have the infrastructure to make them Hidden Gems.

The Pull Forward: From Frameworks to Action

The frameworks in this article are not just theoretical exercises. They are diagnostic tools. Use the Value Leakage Model this week to audit your most important partnership. Use the Value Capture Matrix to start a conversation with your team about where you are placing your bets.

The hard truth is that most partnerships fail quietly from neglect, not spectacularly in a press release. They die from a thousand small cuts of misalignment, friction, and ambiguity. But by shifting your mindset from pure extraction to value creation, and by using these frameworks to systematically de-risk your initiatives, you move from playing the lottery to building a system.

This is the foundational work. It’s the diagnosis before the treatment. In a future post, I will lay out a concrete, 90-day playbook to operationalize this strategy and build a high-ROI partnership engine. But the work of value capture begins now, with the questions you ask and the systems you build.

About Line of Sight

Line of Sight breaks down how AI, strategy, and revenue growth architecture turn complexity into leverage. Written by Kyle Kelly for VP and C-suite GTM, product, and partnership leaders. Join Strategic Operators and read more at lineofsight.io

References

1. Cagan, Marty. “Value Creation vs. Value Capture.” Silicon Valley Product Group, 21 Apr. 2013.

Meta is becoming a power trader. Microsoft is restarting nuclear reactors. Amazon is building its own energy infrastructure. These aren’t energy side projects. They’re GTM plays.

I spent the last three years at Aurora Solar (B2B SaaS) in the thick of the energy landscape leading partnerships & strategic growth. I built our financing partner ecosystem from zero to millions, scaled strategic partnerships that drove customer acquisition, and executed strategic acquisitions to expand marketplace capabilities.

When I see Big Tech vertically integrating into energy markets, I recognize the pattern: uncomfortable bets that expand into non-obvious categories. This is the same build, buy, partner playbook I ran scaling cross-functional teams across product, sales, marketing, and customer success.

By the end of the read, you will understand the energy bottleneck, the competitive gap with China, and why the Grid Wars are the new platform wars.

The AI race isn’t about models anymore. It’s about who can secure the power to run them.

The AI race is shifting.

Models are improving, but the constraint is no longer architecture. It is electricity.

By 2030, US data centers will consume over 800 terawatt-hours (TWh) of electricity annually. That’s enough to power 75 million homes, more than half of all households in America [1].

For the first time in decades, the grid feels the strain. And the companies building the future of AI are discovering that the scarcest resource is not talent or GPUs. It is megawatts.

This isn’t a distant forecast. It’s a demand shock happening now. The US grid, after two decades of flat growth, is not ready.

For founders & investors, the AI bottleneck isn’t just compute. It’s power. And the companies that solve for energy will win the next decade.

1. The Consumer Impact is Already Here

This isn’t an abstract problem for data center operators. It’s showing up on utility bills. Just ask your friends in Northern Virginia.

Wholesale electricity prices near data center hubs are up 138% to 267% in five years [2]. In Northern Virginia, the world’s largest data center market, residents face a $20/month rate hike explicitly tied to this new demand.

The market is sending a clear signal. The cost of power is becoming a material input for AI companies & a new tax on consumers.

2. The Demand Explosion

AI training runs consume as much electricity as 100 US homes use in a year. Inference at scale multiplies that exponentially.

The US grid added zero net capacity growth in the last two decades [1]. AI data centers are projected to 4x their electricity consumption by 2030. The math doesn’t work.

Wait times for new grid connections in Northern Virginia are seven years [1]. For AI companies racing to deploy models, that’s an eternity. Something has to give.

Big Tech isn’t waiting to find out what.

3. The US-China Infrastructure Gap

China operates with a 90% reserve margin on its grid. The US maintains 15% [3].

This is like driving. The US grid is running with the gas light on. China has a full tank & a spare jug in the trunk.

This structural difference creates a massive gap in speed-to-market. A new data center can get power in China in under 6 months. In Northern Virginia, the wait is seven years. For AI companies, that’s an insurmountable lead without private infrastructure.

Timelines vary by province and by state, but the directional gap is real.

4. Big Tech Becomes the Utility

A predictable hierarchy is forming.

• When a resource becomes scarce

• When timelines extend from quarters to years

• And when the market cannot respond fast enough

Large platforms respond by integrating down the stack. Electricity is no exception.

Amazon, Google, Microsoft, and Meta are not just buying power. They are reshaping energy markets. Through multi decade PPAs, nuclear partnerships, SMR investments, and utility agreements, they are building private energy portfolios at a scale we have never seen.

Faced with a systemic bottleneck, Big Tech is executing a new strategy: becoming the energy market itself. They aren’t waiting for the grid.

They’re building their own.

The strategies fall into three categories:

1. Power Trading & Market Making: Meta, Microsoft, & Apple are becoming electricity traders [1]. This lets them sign 10-20 year contracts that unlock new power plant construction, then trade excess capacity as a hedge. It’s vertical integration into energy markets.

2. Direct Energy Infrastructure: Google & Amazon are building their own power sources. SMRs (Small Modular Reactors) are small, factory-built nuclear batteries installed onsite, bypassing the public grid entirely [5] [6]. Amazon also secured 1.9 GW from Talen Energy’s existing nuclear plant through 2042 [11].

3. Utility Partnerships: Microsoft, Google, Amazon, & even steel manufacturer Nucor have signed agreements with Duke Energy to accelerate carbon-free energy options across multiple states [9]. Google added a renewable deal with TotalEnergies for its Ohio data centers [10].

OpenAI’s Stargate Project is the most aggressive: $500 billion across five facilities requiring 7 gigawatts [7]. They’re not adapting to the bottleneck. They’re forcing the market to build around them.

Tesla is selling Megapack battery systems to data centers, targeting a 20 GW/year market [8].

5. The Takeaway

Energy is no longer a utility input. It is becoming a competitive moat.

Companies that secure dedicated power will ship faster, train larger models, and serve inference at lower cost. Those tied to the public grid will face delays measured in years.

We have seen this pattern before. AWS for compute. Azure for enterprise cloud. Meta for mobile. Infrastructure investments that looked unconventional at first and became foundational.

The grid wars will follow the same arc.

This is part of my Line of Sight substack. I write about AI strategy, GTM architecture, and the economics shaping the next decade. Subscribe for more.

References

1. The Electricity Supply Bottleneck on U.S. AI Dominance - CSIS, March 2025

2. How AI Data Centers Are Sending Your Power Bill Soaring - Bloomberg, September 2025

3. AI experts return from China stunned: The U.S. grid is so weak, the race may already be over - Fortune, August 2025

4. Constellation to restart Three Mile Island unit, powering Microsoft - World Nuclear News, September 2024

5. New nuclear clean energy agreement with Kairos Power - Google, October 2024

6. Amazon investing in SMRs to deploy 5 GW by 2039 - American Nuclear Society, October 2024

7. OpenAI’s Stargate Project Will Require Energy to Power a Small Country - Business Insider, October 2025

8. Tesla targets AI data centers with massive Megapack batteries - Tom’s Hardware, November 2025

9. Duke Energy, Amazon, Google, Microsoft execute agreements to accelerate clean energy options - Duke Energy, May 2024

10. TotalEnergies to Supply Renewable Power to Google’s Data Centers - TotalEnergies, November 2025

11. Amazon, Google, Meta and Microsoft go nuclear - Trellis, June 2025

Hi, I’m Kyle Kelly. Each week Line of Sight breaks down how AI, strategy, and revenue growth architecture turn complexity into leverage.

800 million people use ChatGPT every week. Now they can buy things without leaving the chat. Here’s how Stripe and OpenAI just rebuilt commerce.

I spent the last week analyzing their integration and the Agentic Commerce Protocol (ACP). As someone who has spent years in the trenches of e-commerce and logistics, including at Zappos/Amazon, I see this as more than just a new feature. It’s a fundamental re-architecting of how we buy and sell online. By the end of the read, you will understand the stack end-to-end, how you might run your own tiny experiments, and why it matters.

This partnership solves the transaction layer in agentic commerce. Agents can now complete purchases, not just recommend them. That’s a big deal. Here’s why.

Subscribe now

1. What the Partnership Actually Is

On September 29, 2025, Stripe and OpenAI announced that ChatGPT users can now buy products directly within the chat interface, powered by a new open standard called the Agentic Commerce Protocol (ACP) [3]. This isn’t just another checkout button. It’s a new set of rails for AI-driven transactions.

So what does this unlock? The protocol defines a standardized way for AI agents, like ChatGPT, to interact with merchant systems. It breaks the process into three core components: a Product Feed, an Agentic Checkout, and a Delegated Payment mechanism [4].

This means developers can now build applications where an AI agent can discover products from a merchant’s catalog, manage a checkout process rendered within the AI’s interface, and securely handle payments without ever exposing the user’s raw credit card details. It’s a conversational, programmatic approach to commerce.

2. I Analyzed the Integration from an Operator’s Lens. Here’s the Reality.

The live integration is currently limited to approved partners like Etsy and Shopify, so I couldn’t test it hands-on. Instead, I did something more useful: I spent three weeks since OpenAI Dev Day analyzing the complete technical documentation from an operator’s perspective.

The question I wanted to answer: What would it actually take to get this running in production?

The process involves three key implementation phases: integrating the product feed, building the checkout endpoint, and handling the new payment tokens. The product feed requires merchants to provide structured data in near real-time, with updates as frequent as every 15 minutes. This is a significant operational lift compared to the daily or hourly updates common today, if you know you know.

What surprised me most was the architecture. OpenAI is explicitly not the merchant of record. The merchant retains full control over payment processing, fraud detection, and order validation. The Agentic Commerce Protocol is also open source, a strategic move that has already prompted PayPal to adopt it rather than build a competing standard [5].

Subscribe now

This architecture is brilliant. It allows merchants to leverage their existing infrastructure and compliance while tapping into a massive new distribution channel. The checkout UI is in ChatGPT, but the transaction logic and customer data remain with the merchant.

3. The Business Model Implications

How does this change the flow of money? For OpenAI, it creates a new, potentially massive revenue stream. The company will take a transaction fee, determined by confidential agreements with partners like Etsy and Shopify [6] who have early access. With 800 million weekly users, even a small fee on a fraction of their commercial-intent queries could generate billions.

For developers and merchants, the return on investment appears immediate. My analysis suggests that even a conservative 5% lift in conversion rate would generate a net gain of approximately $3,300 per month on a baseline of 1,000 orders, easily covering the estimated infrastructure costs. The payback period is effectively zero.

This model also shifts the competitive landscape. It’s no longer just about having the best website or app. It’s about having the best data and the most seamless integration with the AI agents that will increasingly guide consumer purchasing.

4. The Strategic Chess Move

Why did Stripe and OpenAI make this move? For Stripe, I believe it’s about cementing its position as the economic infrastructure for the AI era. By co-authoring the protocol and being the first to offer a compatible payment primitive, the Shared Payment Token (SPT), Stripe embeds itself at the heart of the next wave of commerce. With 78% of the Forbes AI 50 that accept payments already using their platform, this move creates a powerful network effect [1].

For OpenAI, it’s about moving beyond information retrieval to action. ChatGPT becomes a tool that not only answers questions but gets things done. This deepens user engagement and unlocks a transactional revenue model, reducing reliance on subscriptions and API usage alone.

Who wins? The user gets a frictionless experience. The merchant gets a new, high-intent sales channel. And the payment processors who adapt fastest, like Stripe and now PayPal, position themselves as the tollbooths for the agentic economy.

5. How You Can Benefit from This

This partnership creates distinct opportunities for different players in the ecosystem.

For E-commerce Operators: The most immediate use case is AI-powered checkout optimization. By integrating your product feed, you can reduce purchase friction and potentially see a significant conversion lift. Start by auditing your product data quality and inventory sync frequency; the 15-minute update requirement is a high bar.

For SaaS Founders: Think about how you can leverage this for your own billing. Can you build an AI agent that handles subscription management, answers billing questions, and even processes payments for plan upgrades conversationally? The underlying protocol is not limited to physical goods.

For Developers: This is a greenfield opportunity. You can build AI agents that automate complex purchasing workflows, from corporate procurement to personal shopping. The open-source nature of ACP means you can build on top of this standard without being locked into a single vendor.

Subscribe now

6. What This Means for Agentic Commerce

This partnership is the starting gun for agentic commerce. As I wrote previously in “Agentic Commerce Will Break the Ad Business,” the real disruption comes when AI agents can not only discover but also transact on behalf of users. The Stripe-OpenAI collaboration provides the critical transaction layer that was missing.

The market opportunity is staggering. McKinsey projects that agentic commerce could orchestrate up to $1 trillion in the US B2C retail market alone by 2030, with global projections reaching as high as $5 trillion [7]. This isn’t a niche market; it’s the future of how we interact with the digital economy.

This shift raises profound questions. Does this move power to the agent platforms or to the merchants? Initially, it seems to benefit both. But over time, as users come to trust their AI agents, the platform that controls the agent may become the new gatekeeper of commerce, a position once held by search engines and social media.

7. The Future is Conversational

The key takeaway is this: the internet is becoming conversational, and commerce is following suit. The Stripe-OpenAI partnership has laid the rails for this new economy.

My prediction: By Q4 2027, 25% of single-item online purchases will start in an AI chat interface. The companies that will win are not those who treat this as another sales channel, but those who re-architect their business around the new reality of agent-mediated transactions.

What do you think? Are you preparing your saas business or product catalog for agentic commerce? Drop a comment below or DM me, I’d love to hear how you’re thinking about this shift.

References

• [1] Stripe. (2025, February 27). Stripe’s total payment volume reaches $1.4T, fueled by long-standing investments in AI. Stripe Newsroom.

• [2] Bellan, R. (2025, October 6). Sam Altman says ChatGPT has hit 800M weekly active users. TechCrunch.

• [3] Stripe. (2025, September 29). Stripe powers Instant Checkout in ChatGPT and releases Agentic Commerce Protocol codeveloped with OpenAI. Stripe Newsroom.

• [4] OpenAI. Get Started. OpenAI Developers.

• [5] PayPal. (2025, October 28). OpenAI and PayPal Team Up to Power Instant Checkout and Agentic Commerce in ChatGPT. PayPal Newsroom.

• [6] Capoot, A. (2025, September 29). Etsy pops 16% as OpenAI announces ChatGPT Instant Checkout for the shopping site. CNBC.

• [7] Schumacher, K., & Roberts, R. (2025, October 17). The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants. McKinsey.

Two weeks ago, Shopify reported Q3 results that deserve a closer look. When a company re-accelerates growth to 32% while posting 18% free cash flow margins, expands its addressable market 18-fold through product innovation, and reports 7x AI traffic growth in ten months: that’s not luck. That’s a playbook.

What struck me most was the strategic clarity: Shopify has evolved from “website builder for entrepreneurs” to “operating system for all commerce,” and they’re now laying the rails for agentic commerce while everyone else is still figuring out what that means.

I dug into their Q3 ‘25 investor deck and 10-Q, then discussed the findings with ecommerce colleagues I’ve worked alongside for over a decade. Here’s what the data reveals about three compounding flywheels that could define the next decade of commerce.

Subscribe now

Shopify’s Q3 2025 results, announced on November 4th, paint a clear picture of a company hitting its stride with renewed vigor and discipline. In a quarter that saw both revenue and Gross Merchandise Volume (GMV) accelerate to 32% year-over-year growth, the e-commerce giant also posted an impressive 18% free cash flow (FCF) margin [1].

This isn’t just a story of post-pandemic normalization; it’s a narrative of deliberate, strategic execution transforming Shopify from a high-growth SaaS tool into a profitable, indispensable operating system for all of commerce.

After analyzing the company’s investor presentation and the market’s response over the past two weeks, a strong thesis emerges: Shopify’s current success and future trajectory are powered by three compounding flywheels. These are relentless product-led Total Addressable Market (TAM) expansion, deepening payment penetration that boosts take rates and margins, and a forward-looking pivot to become the essential infrastructure for the nascent world of AI-driven agentic commerce.

This combination is not just driving growth; it’s cementing Shopify’s position as the foundational layer for the next era of retail.

The Core Thesis: A Unified Platform for All Commerce

Before diving into the numbers, it’s crucial to understand Shopify’s core investment thesis, which they lay out with simple clarity. It’s a four-part statement that acts as the north star for their entire strategy.

This thesis is a powerful declaration of intent. Shopify is no longer just a tool for entrepreneurship, it aims to be the platform for all of commerce.

“As our merchants do better, Shopify does better.”

This final point is the fundamental principle of alignment that makes their business model so resilient and scalable. Every new feature, every integration, and every strategic decision can be traced back to this core idea of shared success.

The Re-acceleration Machine

For years, the question for high-growth SaaS has been, “What happens when growth slows?” Shopify’s answer is to re-accelerate. The company’s Q3 performance marks a significant uptick, with revenue hitting $2.8 billion, the strongest growth rate in recent quarters.

This isn’t a single-quarter anomaly but the result of sustained momentum. As the chart below illustrates, the 32% YoY revenue growth demonstrates a clear acceleration from the low-to-mid-twenties growth rates of previous quarters.

The revenue mix, with 75% from Merchant Solutions (which includes payments, fulfillment, and capital) and 25% from high-margin Subscription Solutions, shows that Shopify is successfully capturing more value as its merchants grow. This is the “shared success” model in action.

The acceleration is particularly notable because it comes alongside improved unit economics. Every new merchant added to the platform has a higher lifetime value than those from previous cohorts, driven by increased adoption of Merchant Solutions and multi-channel selling. This is growth that compounds, not growth that burns capital.

The Margin Expansion Engine: Payments Penetration

Perhaps the most critical lever for Shopify’s financial success is the continued adoption of Shopify Payments and its consumer-facing counterpart, Shop Pay. For a platform business, increasing the take rate, the percentage of GMV captured as revenue, is paramount. Shopify Payments is the primary vehicle for this.

The Numbers Tell the Story

In Q3 2025, Shopify Payments penetration reached 65% of GMV, up from 58% in 2023 and just 45% in 2020. This steady climb directly translates to higher-margin revenue and greater free cash flow. The Gross Payments Volume (GPV) for the quarter hit $60 billion, showcasing the immense scale of this operation.

With hundreds of millions of buyers opted-in and a staggering $280 billion in cumulative GMV since its launch, Shop Pay creates a frictionless checkout experience. The Q3 growth numbers tell the story: Shop Pay GMV grew 67% year-over-year, accelerating from 57% in Q1 and 65% in Q2.

This isn’t just adoption; it’s acceleration of adoption.

The Data Moat

The strategic brilliance here is that Shopify Payments isn’t just a revenue line item; it’s a data moat. Every transaction processed through Shopify Payments generates insights that feed back into:

• Fraud detection

• Credit underwriting for Shopify Capital

• Personalized recommendations

This data advantage compounds over time, making the platform more valuable to merchants and harder for competitors to replicate.

The 18x TAM Expansion Playbook

Shopify’s growth story is also one of remarkable market expansion, driven not by acquisitions but by relentless product innovation. The company has systematically expanded its TAM from $46 billion in 2015 to an estimated $849 billion today [2].

This was achieved by evolving from a simple website builder into a “unified operating system for commerce.”

This slide is the visual manifestation of their strategy. Shopify is the central hub connecting a merchant’s online store to physical retail, social media channels, marketplaces, and B2B operations.

Over 90% of merchants have installed two or more channels.

This isn’t theoretical strategy; it’s reality. This multi-channel integration makes the platform incredibly sticky. A merchant who sells online, in-store via POS, on Instagram, and through Amazon, all managed through Shopify, faces enormous switching costs. The unified inventory, customer data, and analytics across all these channels become mission-critical infrastructure.

Two Growth Engines: Offline and B2B

Two key areas of this expansion are Offline and B2B commerce, which are now significant growth drivers and represent massive TAM expansion opportunities.

The offline story is particularly compelling because it represents a complete reversal of the traditional e-commerce narrative. Instead of physical retail dying, Shopify is enabling a unified commerce model where online and offline reinforce each other. A customer can browse online, buy in-store, and return via mail, all tracked in a single system.

This is the future of retail, and Shopify is building the infrastructure for it.

With only 2% penetration of its current Serviceable Addressable Market (SAM) of $404 billion, the runway remains incredibly long.

The New Frontier: Agentic Commerce

The most forward-looking element of Shopify’s strategy is its aggressive push into agentic commerce. The company recognizes that the next paradigm shift in online shopping will move from search bars and websites to conversations with AI agents.

Shopify is positioning itself to be the rails on which this new world runs.

A Three-Layered AI Strategy

Shopify’s AI strategy is a comprehensive, three-layered approach that demonstrates unusual strategic clarity:

Layer 1: Agentic Commerce Infrastructure

Shopify has built three foundational products that allow any AI agent to sell products from Shopify merchants:

• Shopify Catalog provides real-time data on hundreds of millions of products

• Universal Cart allows shoppers to add items from multiple stores into a single cart

• Checkout Kit embeds Shopify’s battle-tested checkout directly into AI agent experiences

This isn’t vaporware; Shopify has already partnered with OpenAI (ChatGPT), Microsoft Copilot, and Perplexity to enable in-chat shopping.

Layer 2: AI Tools for Merchants

Shopify is embedding AI throughout the merchant experience. Sidekick, their personalized AI assistant, helps merchants make smarter decisions, from product descriptions to marketing campaigns. The adoption here is impressive: 750,000+ merchants used Sidekick for the first time in Q3 alone.

This is legit because it demonstrates that AI isn’t just a feature for power users; it’s becoming table stakes for all merchants.

Layer 3: Internal AI Tools

Shopify is using AI to build better products. Scout, their internal AI assistant, has indexed hundreds of millions of merchant feedback points, allowing employees to quickly surface insights and make data-driven decisions.

This creates a flywheel: better products attract more merchants, which generates more data, which trains better AI models, which enables even better products.

The Early Data is Striking

Shopify reported that traffic from AI tools to its merchants’ stores was up 7x since the start of the year, while purchases attributed to AI-powered search had exploded by 11x [3].

The fact that orders are growing significantly faster than traffic suggests that AI-driven recommendations are leading to higher conversion rates.

This is the critical insight: AI shopping isn’t just shifting existing demand; it’s creating better matches between buyers and products, which increases overall conversion and GMV.

As President Harley Finkelstein stated, “AI is not just a feature at Shopify. It is central to our engine that powers everything we build” [3]. More importantly, he noted that Shopify’s goal is to “power product discovery for ALL agents, making this the standard across the internet.”

This is an audacious bet, but it’s grounded in a real competitive advantage. Shopify has the product catalog, the checkout infrastructure, and the merchant relationships. If AI shopping takes off, and early data suggests it will, Shopify is positioned to be the Stripe of agentic commerce.

A New Era of Disciplined Growth

Underpinning this strategic execution is a newfound operational discipline. After a period of heavy investment and the painful decision to exit logistics, Shopify has demonstrated significant operating leverage.

Adjusted operating expenses as a percentage of revenue have fallen from 60% in early 2023 to just 32% in Q3 2025. This efficiency has unlocked consistent and growing profitability, with the company posting nine consecutive quarters of double-digit free cash flow margins.

This chart shows the inverse correlation between operating expenses as a percentage of revenue (a falling line is good) and the free cash flow margin (a rising line is good). This is the hallmark of a mature, scalable platform that can grow efficiently while still investing in future opportunities.

The company is guiding to Q4 operating expenses of 30-31% of revenue, suggesting this discipline will continue.

The Efficiency Story

The breakdown of operating expenses tells an important story:

• Sales and marketing has fallen from 16% of revenue in Q3 2023 to 14% in Q3 2025

• R&D has dropped from 14% to 10%

This isn’t cost-cutting; it’s operating leverage. The platform is becoming more efficient at acquiring and serving merchants, which allows Shopify to invest more in absolute dollar terms while spending less as a percentage of revenue.

Why This Matters

Shopify’s Q3 2025 performance is more than just a strong earnings report; it’s a blueprint for the next generation of platform companies. It demonstrates that it’s possible to achieve venture-scale growth, expand into new markets, and invest in future-defining technologies like AI, all while maintaining financial discipline and generating significant free cash flow.

The company has successfully navigated the transition from a simple tool for entrepreneurs to a complex, multi-product, global infrastructure for commerce.

The three flywheels (TAM expansion, payments penetration, and the AI commerce layer) are not independent; they are deeply interconnected and self-reinforcing.

More products attract more merchants, who bring more GMV, which increases payments volume and generates the data needed to build a world-class AI commerce engine. This is the powerful, compounding loop that will define Shopify’s next decade of growth.

The Timing is Everything

What makes this particularly compelling is the timing:

• E-commerce still represents only 15-21% of total retail sales globally

• The shift to unified commerce (online + offline) is just beginning

• Agentic commerce is in its infancy

Shopify is positioned at the intersection of all three trends, with the infrastructure, merchant relationships, and financial resources to capitalize on each.

The Playbook

For investors, operators, and anyone building in the commerce space, Shopify’s playbook offers critical lessons:

1. Align incentives with your customers - Make customer success metrics your primary KPIs

2. Expand your market through product innovation rather than acquisition - Focus on adjacencies that leverage existing infrastructure

3. Build network effects through payments and data - Own the transaction layer to capture data and improve unit economics

4. Invest early in platform shifts like AI - Build infrastructure before the use cases are obvious

Execute on all four, and you might just build the operating system for an entire industry.

References

1. Shopify Inc. (2025, November 4). Shopify Beats Across the Board: Q3 Shows Growth, Efficiency, Brand Wins. Shopify News.

2. Shopify Inc. (2025, November 4 ). Investor Overview Deck - Q3 2025.

3. Perez, S. (2025, November 4 ). Shopify says AI traffic is up 7x since January, AI-driven orders are up 11x. TechCrunch.

4. Kinsta. (2022, June). Shopify Market Share and Usage Statistics.

5. Statista. (2025, September ). U.S. e-commerce platforms market share 2025.

Disclaimer

This article presents analysis and opinion, not investment advice. All data is sourced from publicly available materials and believed to be accurate, but readers should conduct their own due diligence before making any investment decisions. The views expressed are those of the author and do not constitute recommendations to buy or sell any securities.

What if the most valuable real estate on the internet is no longer the top of the search results page? For two decades, Google and Amazon have built empires on that single premise. Billions have been spent optimizing the moment before a click. Agents erase the click entirely. Now, a new model threatens to make it obsolete, challenging the foundational business model of the entire ad-supported internet.

I remember when Amazon’s ad business was just a few small experiments tucked inside the search bar. As Head of Business Development at Zappos, an Amazon company, I had a front-row seat to how those experiments became a $60 billion revenue engine [1], built entirely on monetizing the tiny moments of indecision between a customer’s search and their final purchase. But its biggest threat isn’t Walmart; it’s the rise of a new paradigm: agentic commerce.

The recent legal clash between Amazon and Perplexity AI is the first visible battle in a new war for digital demand [2]. While Amazon claims this is a “security issue,” it is fundamentally an economic one. And while Perplexity is the catalyst, the true disruption is the underlying shift to agentic commerce itself.

The Digital Shopping Mall vs. The Personal Shopper

Think of Amazon’s marketplace as the world’s largest shopping mall. Its advertising business is the revenue it generates from premium shelf space. Brands pay top dollar for the end caps, the eye-level placements, and the prominent displays, all to influence your path through the store. This “digital shelf space” has been incredibly effective.

Now, imagine you hire a professional personal shopper, an expert who knows your style, your budget, and the best products across every store in every mall. This shopper doesn’t browse; they execute. They go directly to the best item, ignoring the flashy displays and paid placements. That’s agentic commerce.

This is what I call “quiet commerce.” For transactional purchases, it bypasses the entire discovery-and-influence model. While users will still enjoy browsing for discovery-driven shopping, the emotional, status-driven side of commerce, for goal-oriented, utilitarian tasks, AI is built to eliminate indecision. This is especially true for high-substitution products like cables or appliances, whereas categories like fashion and luxury goods will still rely on brand and emotion. This shift fundamentally changes the customer relationship, moving it from the platform to the agent, a misalignment detailed in the table below.

The Agentic Commerce Value Chain

To understand the depth of this shift, it’s helpful to map the Agentic Commerce value chain, which reveals where value is created, where it’s captured, and where the biggest disruptions will occur.

Every disruptive shift has a value chain. Agentic commerce is no different.

Layer 1: Infrastructure

This layer provides the raw materials for agentic commerce through two parallel tracks: Data Infrastructure and AI Infrastructure.

Data infrastructure includes product catalogs, pricing, and reviews from Shopify, Walmart, and Amazon. AI infrastructure includes LLMs from OpenAI, Anthropic, and Google, vector databases from Pinecone and Weaviate for semantic search, and cloud compute from AWS and Azure.

Companies here capture value through API calls and compute usage. The layer is stable and oligopolistic with high barriers to entry.

Layer 2: Data Access & Legal

This is the critical chokepoint. Agents need product data, pricing, inventory, and reviews to make recommendations. That data comes from public APIs, web scraping, and platform agreements.

Platforms control the data. Amazon can restrict API access. Walmart can block crawlers. Shopify can change terms of service. This layer determines whether agents can operate at all.

Amazon is already weaponizing this chokepoint. The lawsuit against Perplexity isn’t just about security. It’s about protecting the ad moat by controlling who can access product data. If platforms cut off data access, agentic commerce collapses.

This is where the battle will be fought.

Layer 3: Agent Platform

This is where the “brains” reside. Consumer platforms like Perplexity and ChatGPT provide the user interface and orchestrate the tools and data sources needed to fulfill requests.

Some agents are platform-specific (Amazon Rufus, Shopify Shop app). Others are platform-agnostic (Perplexity), searching across the entire web to recommend products from any retailer.

Value is captured through subscriptions (Perplexity Pro) or driving sales within platforms (Rufus). The $60B opportunity lies in becoming the trusted agent users turn to first, displacing Google and Amazon as the starting point for product search.

Platform-agnostic agents are aligned with users, not retailers. That’s what makes them disruptive.

Layer 4: Transaction & Platform Interface

Once a decision is made, this layer executes the purchase through two paths.

Transaction Path (Stable)

Payment processors like Stripe and PayPal handle transactions. Order management companies like ShipStation and Flexport manage logistics. Customer service platforms like Zendesk handle post-purchase support.

These companies capture value through transaction fees and subscriptions. This layer is stable. Payments and logistics work the same regardless of where the customer came from.

Platform Interface Path (Disrupted)

This is the traditional e-commerce layer: platforms like Amazon and Walmart, and marketplaces like Google Shopping and eBay.

In agentic commerce, this layer is being disintermediated. Instead of being the primary destination, it becomes another data source for agents to query. Value capture is threatened as agents bypass advertising and drive purchases based on user preference, not paid placement.

Amazon’s $60B ad business is uniquely vulnerable. The $500B retail media market depends on controlling the customer journey. If agents take over that journey, first-party platform data loses value.

The New Unit Economics of Search

The disruption goes deeper than just bypassing ads. The underlying economics are shifting from cost-per-query (the price of a search) to cost-per-answer (the price of a solution).

While AI inference costs remain high and the economics are still unproven, the efficiency of providing a direct answer versus a list of links presents a new model.

This hinges on the agent producing superior recommendations; if trust collapses due to mediocre or biased results, the model fails. But an agent that successfully eliminates the user’s costly exploration phase holds a fundamental advantage.

How fast will this shift happen? Platform transitions are notoriously difficult to predict, but history suggests they accelerate faster than incumbents expect. Mobile shopping went from 10% to 50% of e-commerce in five years. Voice assistants captured 20% of smart home purchases within three years of launch. Even a 10-15 percent shift to agentic commerce by 2030 would represent a $50B-$75B loss for retail media networks, and that assumes platforms can maintain their current share of non-agentic commerce, which is not guaranteed.

This efficiency threatens the entire retail media network market, a sector projected to be worth between $300 billion and $500 billion by 2030, depending on the source [4].

This market is the new gold rush, but its value is predicated on controlling the customer’s journey.

If AI agents become the new front door to commerce, the value of that first-party platform data plummets.

The Innovator’s Dilemma: Protecting the Ad Moat

Amazon is facing a classic innovator’s dilemma. It has its own AI agent, Rufus, but Rufus is an employee of the mall, not your personal shopper.

It will help you find things within Amazon’s walls, but it will never tell you there’s a better deal at Target. This creates a crucial opening for independent, user-aligned agents like Perplexity to win the user’s trust.

Amazon’s lawsuit is an attempt to fortify its ad moat. But Amazon isn’t defenseless. It can, and likely will, retaliate by blocking crawlers, degrading data access, and bundling its own agent (I predict we see this one Rufus 2.0). It also holds the trump card in the supply chain, pricing, and the customer relationship through Prime. Its ad business, however, built on discovery, remains uniquely vulnerable to this new, more efficient model of commerce.

The Counterargument

Of course, this transition is not without its own pitfalls. History suggests that today’s disruptive agents could become tomorrow’s platforms, replicating the same walled gardens they claim to dismantle. This assumes agents have perfect data access, which isn’t guaranteed in a world of walled gardens and API restrictions. Advertising has also survived every platform shift, and retail media networks will adapt by shifting to outcome-based pricing or integrating their own agents.

But the core economic shift remains: agents are fundamentally aligned with the user, not the platform.

Even if agents are imperfect, the incentives have already shifted and that part will not unwind. That misalignment is what makes this disruption different. Even if agents become new gatekeepers, they will compete on the quality of their user-aligned recommendations, a stark contrast to today’s ad-driven platforms

Why This Matters: The Great Ad Unbundling

The threat is most acute for search-based advertising like Google’s, which is also built on being the middleman for discovery. For social platforms like Meta, where ads are driven by entertainment and attention rather than intent, the dynamic is different. Still, for a significant portion of the digital ad ecosystem, agentic commerce unbundles discovery from the platform.

For two decades, the internet has operated on a simple premise: aggregate demand and sell access to it. AI agents disaggregate that demand, handing power back to the individual user. This is the first major legal test, but it won’t be the last. The future of commerce won’t be about ads; it will be about agents.

The key questions for every leader are no longer just about ad spend and ROAS, but about:

• Agent Strategy: How will your business interact with user-aligned agents? Can you afford to be blocked?

• Customer Relationship: When the primary relationship is with an AI, how do you build brand loyalty?

• Data Moats: Is your data advantage defensible when agents can access and compare information across the entire internet in real-time?

The companies that win next decade will be those that earn the right to act on behalf of the user.

This is part of my Line of Sight newsletter. Subscribe for more operator frameworks, analysis, and GTM strategy.

References

• [1] Amazon Quarterly Results. EDGAR SEC Filings. 2025

• [2] Amazon vs Perplexity: Welcome To The Battle For The Future Of Commerce. Forbes. Nov 5, 2025

• [3] Retail Media Networks Market Size and Share Report. Grandview Research.

• [4] The agentic commerce opportunity. McKinsey. Oct 17, 2025

Apple’s partnership with Google’s Gemini isn’t a headline about AI. It’s a masterclass in sequencing strategy.

When a company this dominant changes its build-buy-partner calculus, it signals a platform shift big enough t…

Read more

Cohort analysis is supposed to be descriptive, not predictive.
But if you’re the kind of operator who looks for signal in the noise, hear me out.

When you segment by behavior instead of time, patterns emerge early.
You sta…

Read more

You all know I’m obsessed with how partnerships actually create value.
Not the logos, co-marketing, or handshakes, but the systems that turn collaboration into compounding revenue.

I’ve led teams that built partner ecosys…

Read more

AI protocols are still evolving, but the early signs are clear: standardization is becoming strategy. Whether these layers converge or fragment, they’re already changing how software products distribute and compete.

Read more

This week, I analyzed how ChatGPT, Midjourney, and Loom reached massive scale without traditional marketing budgets. While most companies still treat product and distribution as separate functions, I’ve been thinking about what…

Read more

This week, I analyzed Intuit’s recent Investor Day presentation. While companies announce new AI releases every week, I’ve been thinking about what separates announcements from execution. Everyone talks about copilots and feat…

Read more

This week, I’ve been thinking about distribution, not as a marketing function but as the ultimate power layer of AI. Every company talks about models, but the real game is reach, control, and compounding access. I’ve watched this same…

Read more

Subscribe now

Leadership frameworks aren’t invented in boardrooms. I learned mine at 5 AM on the Schuylkill River.

While my classmates slept, I was learning how rhythm, feedback, and shared struggle compound into performance. Rowing taught me what every operator eventually realizes: what moves a boat forward under pressure is the same discipline that drives organizations through growth and chaos.

Here are the four rules every operator should know.

My Leadership Manifesto:

1. Operating Rhythm > Coordination

2. Feedback in the Moment Scales Faster

3. Shared Struggle Builds Culture

4. Adaptability Is the Real Advantage

1. Operating Rhythm > Coordination

There’s a difference between eight rowers moving at the same time and eight rowers moving as one. The first time it clicked, the shell lifted out of the water. Effort turned effortless.

Our coach called it collective rhythm. Synchronization is just hitting deadlines together. Rhythm is a shared flow where output compounds beyond any one person.

Operator lesson: The best teams don’t just align on goals; they move in rhythm. At Zappos, a seven-person team in flow outperformed groups of twenty-five operating in coordination. In deals or product launches, rhythm is when finance, product, and sales stop defending silos and start pulling together.

2. Feedback in the Moment Scales Faster

“Hey 4-seat, watch the blade sky! Hands level at the catch.”

I was that 4-seat that day. One quick cue from our coxswain and within a few strokes the boat was flying again.

Our U.S. National Team and Collegiate Hall of Fame coach had one rule: he called an issue once, just once. It wasn’t harshness. It was preparation for race conditions where real-time correction determines whether you medal or miss the podium. It was about fixing it while you can impact the race.

In business: quarterly or annual feedback cycles are malpractice. Teams need micro-corrections that compound. In partnerships, in M&A diligence, in product roadmaps, delayed feedback means compounding errors.

When Zoom’s usage exploded 30x during COVID, engineering didn’t stick to quarterly plans. They implemented daily capacity reviews, hourly performance monitoring, and real-time resource reallocation. Speed of learning beat perfection of planning.

Operator move: build feedback loops that move faster than your competition’s decision cycles. If your industry adjusts quarterly, you need monthly cycles. If they adjust monthly, you need weekly recalibration. Fix things while you’re in the arena.

“The frequency of feedback is directly proportional to the rate of skill acquisition. Research shows that immediate feedback improves performance up to 40% faster than delayed feedback.” — Dr. K. Anders Ericsson

3. Shared Struggle Builds Culture

A 2,000-meter race is seven minutes of anguish. Frozen hands, burning lungs, legs on fire. Yet those are the memories rowers talk about decades later. Struggle forged bonds no offsite ever could.

Operator lesson: shared adversity creates trust faster than comfort ever will. In the boardroom, the champagne toast on close day doesn’t build culture. The grind of integration does. The hardest projects, the messiest restructurings, and the pivots under pressure are what forge real teams.

When I’ve deliberately assigned “stretch projects” that felt uncomfortable, teams didn’t just deliver, they came out bonded and more resilient.

4. Adaptability Is the Real Advantage

Lineups changed constantly. One week I was starboard, the next port. Bow pair could move to the engine room overnight. Constant reconfiguration forced adaptability, not as a value but as survival.

Operator lesson: change is no longer cyclical; it’s constant. Deals collapse, markets shift, priorities reset overnight. The leaders who win aren’t the ones with the most rigid plan—they’re the ones who treat reconfiguration as normal.

In M&A: the best integrations succeed because teams expect to reconfigure midstream. They adapt, recalibrate, and keep the boat moving.

McKinsey found that adaptable teams outperform others by 24% during major change. PwC found that adaptable organizations are three times more likely to become industry leaders.

“Research across 150 organizations found that teams with high adaptability scores outperformed their less adaptable counterparts by 24% during periods of significant change.” — McKinsey, Organizational Resilience Study (2023)

The Compound Effect

These four rules operate like the mechanics of a racing shell.

• Synchronization without feedback leads to coordinated failure.

• Shared struggle without pressure doesn’t build capability.

• Feedback without synchronization creates chaos.

When all four combine, something remarkable happens. Teams move faster than their individual capabilities suggest. They anticipate problems before they appear. They know when to push and when to adjust.

Rowing taught me that leadership isn’t about shouting from the stern. It’s about building systems where excellence emerges from the crew.

Final Pull

This is part of LineofSight.io, where I turn AI noise into operator strategy for non-technical GTM leaders.

Subscribe for weekly frameworks on building leverage, structuring deals, and scaling teams that compound.

Subscribe now

TLDR: Your best acquisition target may already be in your partner ecosystem. The integrations, joint customers, and roadmap overlap you see every day are not just partnership signals. They are M&A intelligence hiding in plain sight.

The Blind Spot in Corporate Development

Your next acquisition target is not buried in a banker’s pitch. It is already inside your ecosystem.

Snowflake underscored this at its 2025 Summit. The company highlighted how partner integrations now drive the bulk of usage across its Data Cloud. Partners are not just feature extensions — they are revenue engines. And when Snowflake acquired Streamlit (data apps) and Neeva (AI search), it wasn’t a surprise. Both had already been battle-tested inside the ecosystem.

Most corporate development teams miss this. They spend months scanning banker decks, attending demo days, and waiting for processes to start. Meanwhile, the highest-signal targets are already embedded in their product stack, serving joint customers, and quietly proving value every quarter.

The blind spot is not a lack of capital or vision. It is treating partnerships only as distribution, instead of what they truly are: live diligence processes.

Partnerships as Diligence

Partnerships have evolved. They are not just about co-marketing or channel reach. They are the ultimate diligence process.

Through a strong partner integration, operators can see:

• Real usage data

• Joint customer outcomes

• Cultural alignment in working teams

• Roadmap fit against strategic priorities

That is intelligence you will never find in a CIM.

Partnerships as Due Diligence

The strategic shift is simple. Partnerships are not just about distribution anymore. They are the ultimate diligence process.

Through a strong partner integration you can see:

• Real usage data

• Joint customer outcomes

• Cultural alignment in working teams

• Roadmap fit against your strategic priorities

This is intelligence that no banker deck will provide.

Case Studies: When Partnerships Became Acquisitions

• Salesforce + MuleSoft ($6.5B, 2018)
  Three years of partnership powering 1,000+ integrations. Premium of 36 percent, but de-risked by usage data.

• Dropbox + DocSend ($165M, 2021)
  Two years of API integration with 65 percent customer overlap. Integration complexity solved before the deal closed.

• Square + Afterpay ($29B, 2021)
  Eighteen months of payments integration with thousands of merchants. The partnership created a moat competitors could not cross.

These weren’t speculative bets. They were validated partnerships upgraded into acquisitions.

The Operator’s Take

Think of partnerships as extended job interviews. You would never hire a VP without observing performance. Why would you acquire a company without testing how it integrates with your systems, your customers, and your culture?

Traditional M&A relies on projections and promises. Partnership-first M&A relies on performance data. The difference is dramatic.

The Partnership-to-Acquisition Signal Framework

Green Light (Consider Acquisition):

• Partner drives more than 20% of new customer pipeline

• Joint solution is mission critical for shared customers

• Roadmaps align with your 3-year strategy

• Integration costs declining while value creation rises

• Retention rates improve for joint solutions

Yellow Light (Deepen Partnership):

• Strong technical fit but limited market traction

• Cultural alignment but unclear strategic value

• High customer satisfaction but low revenue attribution

• Integration works but monetization not clear

Red Light (Keep at Arm’s Length):

• Strategic priorities diverge

• Integration complexity rising

• Partner’s wins do not translate to your success

• Cultural friction shows up in collaboration

What Strategic Operators Can Track

You do not need to understand APIs or financial modeling to spot potential acquisitions. Track three business metrics:

• Customer retention: Joint solution retention vs standalone products

• Revenue attribution: Growth from partner-sourced deals and upsells

• Support volume: Tickets on integrated features (declining = good)

If these trend positively, you are looking at an acquisition candidate.

Every Partnership Is M&A Intelligence

Traditional process:

• Identify targets through banker outreach

• Run 60–90 days of diligence

• Decide on projections

• Hope integration works

Partnership-first process:

• Partner first to test fit

• Gather 12–24 months of data

• Decide based on proven performance

• Integrate along an already tested path

The Operator’s Checklist

Ask these five questions of every strategic partner:

1. How much of their revenue flows through our ecosystem?

2. Are we solving the same customer problems?

3. Are customers clearly more successful using us together?

4. Are our leadership teams strategically engaged?

5. If a competitor acquired them tomorrow, how badly would it hurt?

These separate “nice to have” partners from acquisition-ready opportunities.

Your Competitive Advantage

The edge is not having more partnerships. It is knowing when one should become an acquisition.

The playbook is straightforward:

• Partner first to test strategic and cultural fit

• Measure relentlessly to build performance data

• Act decisively when the signals align

Every integration is due diligence. Every joint customer is validation. Every quarterly business review is M&A intelligence.

Operators who adopt this mindset consistently find better targets, at better prices, with lower risk. If you are not evaluating your partner ecosystem with an M&A lens, someone else already is.

This is how you find your next acquisition hiding in plain sight.

Subscribe now

This is what I write about at Line of Sight: operator frameworks and playbooks that turn hidden signals into strategic advantage. Subscribe to get them directly.