https://www.exaptationai.com/blog daily 0.75 2026-10-22 https://www.exaptationai.com/blog/exaptation-growth-flywheel monthly 0.5 2026-10-21 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/db50a9ce-5790-4e26-b49a-34a093120729/fig1-amazon-flywheel-3200x1800.png Blog - The Exaptation Growth Flywheel: How New Uses Build Momentum Figure 1. The Amazon flywheel, redrawn. What the Amazon wheel assumes Amazon's flywheel grows by doing more of the same thing, better. Each turn adds products, sellers, and shoppers, but the function of the store stays the same. Selection is still selection, and a warehouse is still a warehouse. That is a strength, not a flaw. It is the reason the Amazon wheel is so easy to understand. But it leaves a question open: what happens when the next turn of growth comes from a new function, not more of the old one? That is the question EGS was built to answer. Amazon's wheel compounds momentum inside one function. An exaptation flywheel compounds by adding new functions, and every new function that sticks can become an asset the next turn can work with. I've argued before that growth is a search problem. Most efficiency programs try to do more with less. The EGS flywheel below does more with more: more functions found in what is already built, by more people, in more settings. The EGS model in brief The EGS model describes how a new function emerges from an existing asset. In plain terms, it has six parts: • Resources: the assets a firm has, along with their properties. • Natural phenomena: how the world works, from physics to markets to social behavior. A resource meeting these produces an action, which is what the resource actually does. • Agents: the people and organizations that use the resource, each with their own goals, needs, and problems. • Designed functions: the use a resource was built or adopted for. It comes from resources and agents interacting in a given context. • Selective pressures: the forces that keep changing those contexts, such as new technology, new competitors, and shifting customer needs. • Emergent functions: new patterns of use that recur, stabilize, get adopted, and spread. These are the exaptations. A flywheel needs parts that can grow, so the model needs one adjustment before it can spin. A firm can build up its asset base, its users, its experiments, its new functions, and the value they create. It can't easily discover more physics, and it can't manufacture market pressure. Natural phenomena and selective pressures are the conditions the wheel turns in. So in the flywheel below, natural phenomena (those arising from the universe, society, markets, technology, etc.) sit inside the first link, as the reason an asset behaves the way it does. Selective pressures sit inside the second link, as the range of settings a firm exposes its assets to. Designed functions play a different role. They are the starter motor. A designed function is what brings the first customers and users to an asset, the way books brought Amazon its first customers. The five links of the exaptation flywheel Here is the loop, written the way Collins asks for it: each part causes more of the next, and each link has a reason. 1. Asset Base: More assets bring more people to work with them. An asset can only attract new users if its properties and functions are visible and described. People can't build on what they can't see. In the Stanford $5 Challenge (EGS Case Study No. 1), a 3-minute class presentation slot was worth nothing until one team saw it as a resource and sold it to a company that wanted to recruit d.school students. At a much larger scale, Nvidia made the parallel structure of its graphics chips usable by outsiders through CUDA, a software platform it gave away for free. Nvidia has reported more than a billion CUDA-capable GPUs in the world, a very large and very legible asset base for others to work on (NVIDIA). 2. Agents and Motivations: More people, in more settings, try more new interactions. Every person or organization brings a goal, need, or problem of their own. So each new user in each new setting is a new pairing of asset and purpose. Newcomers also arrive without the owner's habits about what the asset is for, which is why new agents show up in nearly every case of exaptation we have studied. When Nvidia passed 2 million registered developers, it was seeing about 438,000 CUDA downloads a month (NVIDIA). Each download is another chance for someone to use the same chips in a way Nvidia never planned. 3. Resource-Agent Interactions: More interactions produce more new uses that stick. Most new interactions go nowhere. A few recur, stabilize, get adopted, and spread. If the share that survives holds roughly steady, more attempts produce more survivors, which is why EGS treats discovery as a portfolio of tests rather than one big bet. 4. Value Generation is the Goal: More new uses create more value. Each new function opens new customers and markets on an asset that is already built and paid for. Uber Eats (EGS Case Study No. 3) used the same drivers, routing, payments, and app that transported riders to deliver food instead. Uber's Delivery segment reported $17.2B in revenue for FY2025. 5. Value Speeds up the Second Turn: More value, shared, brings in more people and more assets. This link closes the loop in two ways. First, when a firm leaves value on the table for others, outsiders have a reason to build on its assets, which feeds link 2. As I argued in the 7 Powers post, the best long-run capture is less than 100%, and over-capturing starves the network. Nvidia shows the pull. Its CUDA developer base grew from 1.8 million in 2020 to more than 4.5 million (SlashData). Second, a new use that sticks can become an asset in its own right, with properties of its own, which feeds link 1. Google Search (EGS Case Study No. 4) began as an exaptation of Weblinks, which were designed for navigation. Search then became the base for further layers of growth, and generated about $1.27 trillion in revenue from 2017 through 2025. The Amazon parallel is closest at link 5. Amazon's sellers add selection. In this wheel, outside agents add functions. On both wheels, the outside party ends up carrying a large share of the load. https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/5754cab0-83e7-4f7b-b58f-92dda47cae2d/fig2-exaptation-flywheel-3200x1800.png Blog - The Exaptation Growth Flywheel: How New Uses Build Momentum Figure 2. The exaptation growth flywheel. What speeds the wheel up Amazon didn't stop at four parts. Once the core loop worked, it added Prime and advertising to make each turn faster. An exaptation flywheel has accelerants too. Exaptation AI at every link. This is the part we are building at Exaptation AI: A search engine for every asset's undiscovered functions. AI can help make assets legible at a scale people alone can't reach (link 1), which is where our Latent Asset Review starts. It can propose interactions that a team used to one function wouldn't think of (link 2), and screen candidates before money is spent so more experiments survive (link 3). It can also record each new use as a new asset, so the next turn starts from a larger map (link 5). The hardest link to support is link 4: matching a new resource function with the customers who need it. That is still an open problem, and we're working on it in public. Designed flexibility. Some firms build assets so that outsiders can create functions the owner never designed. CUDA is the example. Nvidia didn't design each research use of its chips. It made the chips programmable and let others find the uses, which speeds the move from link 1 to link 2. It is an example of a General Purpose Technology. Partnerships. A partnership or joint venture can speed three parts of the wheel in one move, because it brings new people, new settings, and new assets at once. The condition is the same as link 5: both sides have to keep enough value to keep building. Spreading costs. Every new function on an existing asset spreads that asset's cost over more uses, which leaves more room to share value with outsiders. This is the closest match to Amazon's lower-cost loop. I treat it as an accelerant rather than the engine, because EGS is a growth strategy, not an efficiency program. What slows the wheel down Collins's doom loop is what happens when a company keeps changing direction. An exaptation flywheel has its own ways to stall: • Taking too much. When the owner captures nearly all the value, outsiders lose their reason to build. A full acquisition can do the same thing, by bringing in the buyer's habits and ending the outside search. • Seeing only the designed function. When an asset's original use carries the revenue and the identity of the business, it becomes hard for anyone inside to see the asset any other way. It’s called Functional Fixedness and this blindness can leave a lot of value on the table. • New uses that hurt existing users. A new function that makes current users worse off turns the value negative, and people pull back. LinkedIn Dating (EGS Case Study No. 6) is our cautionary case: a new use that made the network worse for the professionals it was built for. What we're still testing A flywheel is a claim about cause and effect, and this one rests on two assumptions we are testing. 1) Link 3 assumes that the share of new interactions that stick holds roughly steady as volume grows. If more attempts bring proportionally more noise, volume alone won't make the wheel spin faster, and the quality of the people and settings would matter more than their number. 2) Placing natural phenomena and selective pressures inside the links, rather than as parts of the wheel, is a modeling choice. It fits how firms actually operate, since no firm controls either one. It is still open to challenge, and I'd welcome it. Where to start According to Stone's account, Amazon's executives felt that sketching the flywheel helped them understand their own business after five years of running it. The same exercise works here, one asset at a time. Pick one asset. Ask what it does today, who uses it, and what else it could do for people who aren't using it yet. Then ask which of the five links is weakest. That link is usually where the next turn is stuck. If you'd like to try this on one of your own assets, a Latent Asset Review is a good place to start, or you can book a conversation through our Appointments page. Sources and related reading Sources • Jim Collins, The Flywheel Effect, with excerpts from Good to Great (2001) • Jim Collins, Turning the Flywheel (2019), excerpt in Outlook Business, quoting Brad Stone, The Everything Store (2013) • Marketplace Pulse, Amazon Is Regaining 1P Unit Share (Q1 2026 unit mix and segment growth) • NVIDIA, 2 Million Registered Developers, Countless Breakthroughs • SlashData, Why NVIDIA dominates despite low developer program scores • Michael Parekh, AI Shoulders to the Flywheel (quoting Collins in Vox) • Uber Technologies, Form 10-K for fiscal 2025 (Delivery segment revenue) Related reading on Exaptation AI • The Eighth Power Problem: Helmer's 7 Powers and EGS • Found Fast, Selected Slow: The Substrate Behind Nvidia's Two Exaptations • Thirteen Strategy Frameworks, One Unasked Question • EGS Theory to Business Practice: the six-step search process https://www.exaptationai.com/blog/substrate-behind-two-nvidia-exaptations monthly 0.5 2026-10-08 https://www.exaptationai.com/blog/google-search-a-1-trillion-exaptation monthly 0.5 2026-10-08 https://www.exaptationai.com/blog/exaptation-is-not-the-goal-value-is monthly 0.5 2026-09-21 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/85f0dab8-44c3-451e-868a-4f05b1e2a4dd/ExAI_VG_Flywheel_v0.2_2500x1406_squarespace%5B1%5D.png Blog - Exaptation Is Not the Goal. Value Is. - Make it stand out The Value Generation flywheel in EGS. Five stages, clockwise. Discovery is where EGS operates; the other four run on practices most firms already have. Source: Shah (2024), Exaptation Growth Strategy, Pepperdine Graziadio Business School. https://www.exaptationai.com/blog/thirteen-strategy-frameworks-one-unasked-question monthly 0.5 2026-09-17 https://www.exaptationai.com/blog/helmer-7powers monthly 0.5 2026-09-17 https://www.exaptationai.com/blog/ubereats monthly 0.5 2026-09-28 https://www.exaptationai.com/blog/intro2 monthly 0.5 2026-10-08 https://www.exaptationai.com/blog/intro-1 monthly 0.5 2026-09-17 https://www.exaptationai.com/blog/category/Exaptation+Growth+Strategy+%28EGS%29 monthly 0.5 https://www.exaptationai.com/blog/category/Nvidia monthly 0.5 https://www.exaptationai.com/blog/category/Theory monthly 0.5 https://www.exaptationai.com/blog/category/Strategic+Management monthly 0.5 https://www.exaptationai.com/blog/category/Exaptation+AI+%28ExAI%29 monthly 0.5 https://www.exaptationai.com/blog/category/Foundation monthly 0.5 https://www.exaptationai.com/blog/category/Case+Study monthly 0.5 https://www.exaptationai.com/blog/tag/Google+Search monthly 0.5 https://www.exaptationai.com/blog/tag/Exaptation+Growth+Strategy+%28EGS%29 monthly 0.5 https://www.exaptationai.com/blog/tag/Theory monthly 0.5 https://www.exaptationai.com/blog/tag/Exaptation+AI monthly 0.5 https://www.exaptationai.com/blog/tag/Exaptation monthly 0.5 https://www.exaptationai.com/blog/tag/Value+Generation monthly 0.5 https://www.exaptationai.com/blog/tag/Uber+Eats monthly 0.5 https://www.exaptationai.com/blog/tag/GPU monthly 0.5 https://www.exaptationai.com/blog/tag/Pace+Layers monthly 0.5 https://www.exaptationai.com/blog/tag/Nvidia monthly 0.5 https://www.exaptationai.com/blog/tag/Growth+Flywheel monthly 0.5 https://www.exaptationai.com/blog/tag/Strategic+Management monthly 0.5 https://www.exaptationai.com/blog/tag/Examples monthly 0.5 https://www.exaptationai.com/blog/tag/Case+Study monthly 0.5 https://www.exaptationai.com/home daily 1.0 2026-09-17 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/6ea7a731-716d-4e3e-b962-56081117d6ef/The+Martian-science-the-shit-out-of-this.jpg https://www.exaptationai.com/appointments daily 0.75 2026-09-17 https://www.exaptationai.com/about daily 0.75 2026-09-17 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/d7d7a6e5-431a-41aa-a01c-86bd35a50ba7/ExAI_Banner_3line_1584x396.png https://www.exaptationai.com/exaptation-science daily 0.75 2026-09-17 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/e5524faf-d05a-41a4-8114-458a8a19e9e3/Bird-warmth-Sora-Aug+2026.png https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/5cbc869d-0c13-4d39-854d-f2486814b96c/Bird-Flight-Sora-Aug+2026.png https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/ca4c9ada-5acb-4695-bfc8-861084441079/Feathers-Display-Sora-Aug+2026.png https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/0887322a-d5ab-4dae-a675-7d2e12747778/Refined+Exaptation+Growth+Strategy+Model+%28EGS%29.jpg https://www.exaptationai.com/services daily 0.75 2026-09-17 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/df123bf3-4a94-4e2e-950d-b6882bc3a92f/ExGPT-OSR-landing+image-Aug-21-26.png https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/50caedc8-2984-4999-bc13-81333a912538/OAI-Platform-Screenshot-ExAI-091426.png https://www.exaptationai.com/services-1 daily 0.75 2026-10-02 https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/df123bf3-4a94-4e2e-950d-b6882bc3a92f/ExGPT-OSR-landing+image-Aug-21-26.png https://images.squarespace-cdn.com/content/v1/69bec2421061f03bfd9374ad/50caedc8-2984-4999-bc13-81333a912538/OAI-Platform-Screenshot-ExAI-091426.png