{"id":486769,"date":"2026-02-20T03:40:09","date_gmt":"2026-02-20T03:40:09","guid":{"rendered":"https:\/\/www.newsbeep.com\/ca\/486769\/"},"modified":"2026-02-20T03:40:09","modified_gmt":"2026-02-20T03:40:09","slug":"how-will-openai-compete-benedict-evans","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ca\/486769\/","title":{"rendered":"How will OpenAI compete? \u2014 Benedict Evans"},"content":{"rendered":"

As we all know, OpenAI has been running around trying to join the club, claiming a few months ago to have $1.4tr and 30 gigawatts of compute commitment for the future (with no timeline), while it reported 1.9 gigawatts in use at the end of 2025. Since it doesn\u2019t have the scale of cashflows from existing businesses that the hyperscalers can use, it has so far managed to do this, or at least announce this, with a combination of capital-raising (not all of which has necessarily closed<\/a>) and other peoples balance sheets<\/a> (some of which is also the famous \u2018circular revenue\u2019). <\/p>\n

You can watch plenty of three-hour podcasts discussing all of this, and plenty of people have opinions about TPUs, Nvidia\u2019s product lead, and Oracle\u2019s strategy of borrowing against a declining but cash-generative legacy business to burn its way into the new thing, but how much should the rest of us care? Is this a path to a competitive advantage, or just a seat at the table? <\/p>\n

We don\u2019t really know what AI infrastructure costs will look like in the long term, but it\u2019s quite possible that this turns out like the manufacture of airliners or semiconductors: there are no network effects, but with each generation the process gets more difficult and more expensive, and so those industries have gone from dozens of companies at the cutting edge to just Boeing and Airbus on one hand and TSMC on the other. Semiconductor manufacturing had both Moore\u2019s Law, which everyone has heard of, and Rock\u2019s Law, which most people haven\u2019t: Moore\u2019s Law said that the number of transistors on a chip was doubling every two years, but Rock\u2019s Law said that the cost of a state-of-the-art semiconductor fab was doubling every four years. Maybe generative AI will work the same, with unit costs falling but fixed costs rising to the point that only a handful of companies are able to sustain the investment needed to build competitive models and everyone else is squeezed out.* This oligopoly would presumably have a price equilibrium, though it might be at high or low margins – this might all just be commodity infrastructure sold at marginal cost, especially given some of those at the table will be using their models to power other, much more differentiated businesses. Ask your favourite economist. **<\/p>\n

So, when Sam Altman says he\u2019s raised $100bn or $200bn, and when he says he\u2019d like OpenAI to be building a gigawatt of compute every week (implying something in the order of a trillion dollars of annual capex), it would be easy to laugh at this as \u2018braggawatts\u2019, and apparently people at TSMC once dismissed him as \u2018podcast bro\u2019, but he\u2019s trying to create a self-fulfilling prophecy. He\u2019s trying to get OpenAI, a company with no revenue three years ago, a seat at a table where you\u2019ll probably need to spend couple of hundred billion dollars a year on infrastructure, through force of will. His force of will has turned out to be pretty powerful so far. <\/p>\n

But, again, does that get you anything more than a seat at that table? TSMC isn\u2019t just an oligopolist – it has a de facto monopoly on cutting edge chips – but that gives it little to no leverage or value-capture further up the stack. People built Windows apps, web services and iPhone apps – they don\u2019t build TSMC apps or Intel apps. <\/p>\n

Developers had to build for Windows because it had almost all the users, and users had to buy Windows PCs because it had almost all the developers (a network effect!). But if you invent a brilliant new app or product or service using generative AI, or add it as a feature to an existing product, you use the APIs to call a foundation model running in the cloud and the users don\u2019t know or care what model you used. No-one using Snap cares if it runs on AWS or GCP. When you buy an enterprise SaaS product you don\u2019t care if it uses AWS or Azure.\u00a0And if I do a Google Search and the first match is a product that\u2019s running on Google Cloud, I would never know. <\/p>\n

That doesn\u2019t mean these APIs are interchangeable –\u00a0there are good reasons why AWS, GCP and Azure have very different market shares, and why developers choose each. But the customer doesn\u2019t know or care. Running a cloud doesn\u2019t give you leverage over third part products and services that are further up the stack. <\/p>\n

The difference now, perhaps, is that all of those services were separate silos: there was a common search and discovery layer at the top in Google and Facebook, and common infrastructures at the bottom in the cloud, but all those apps were never connected to each other. Now we have an emerging alphabet soup of standards and protocols for models and websites to talk to each other across ads, e-commerce and some kind of intent and automation (the brief enthusiasm around OpenClaw captured some of this). A website can surface its capabilities so that a subset can just show up in ChatGPT, be it a real estate search or a shopping cart. You\u2019ll tell your agent to look at a recipe on Instagram and order the ingredients on Instacart. Everything can get piped to everything else, and everything can talk to each other! <\/p>\n

Meanwhile, (saying the quiet part out loud), if you could set and control those APIs and manage the flows, that gives you power. Standards have been a basic competitive weapon in every generation of technology – remember Microsoft\u2019s slogan \u2018embrace and extend\u2019. In particular, OpenAI suggests now suggests you\u2019ll use your ChatGPT account as the glue linking all of these together. That\u2019s a network effect! <\/p>\n

I’m not sure about this: I\u2019m not sure that this vision will really work, and if it does, I\u2019m not sure it gives one company dominance. <\/p>\n

First, there\u2019s a recurring fallacy in tech that you can abstract many different complex products into a simple standard interface – you could call this the \u2018widget fallacy\u2019. A decade ago people said \u2018APIs are the new BD\u2019, which was really the same concept, and it mostly failed. This is partly because there\u2019s a huge gap between what looks cool in demos and all of the work and thought in the interaction models and the workflows in the actual product: very quickly you\u2019ll run into an exception case and you\u2019ll need the actual product UI and a human decision. It\u2019s also because the incentives are misaligned: no-one wants to be someone else\u2019s dumb API call, so there\u2019s an inherent tension or trade-off between the distribution that an abstraction layer might give you (Google Shopping, Facebook shopping, and now ChatGPT shopping) and your desire to control the experience and the customer relationship. Remember, after all, that all of Instacart\u2019s profits come from showing ads. <\/p>\n

Of course, this is just speculation – maybe it will all work this time! But the second problem is that if these are all separate systems plugged together by abstracted and automated APIs, is the user or developer locked into any one of them? If apps in the chatbot feed work, and OpenAI uses one standard and Gemini uses another, why stops a developer doing both? This is much less code than making both an iOS and Android app, and anyway, can\u2019t you get the AI to write the code for you? What does that do to developer lock-ins? Meanwhile, yes, maybe I\u2019ll log into all of these services with my OpenAI or Gemini account, but does it necessarily make sense for me to log into Tinder, Zillow and Workday with the same account? And, again, do they want that?<\/p>\n

Hmm. <\/p>\n

As I’ve written this essay, I\u2019ve returned again and again to terms like platform, ecosystem, leverage and network effect. These terms get used a lot in tech, but they have pretty vague meanings. Google Cloud, Apple’s App Store, Amazon Marketplace, and even TikTok are all \u2018platforms\u2019 but they’re all very different. <\/p>\n

Maybe the word I’m really looking for is power. When I was at university, a long time ago now, my medieval history professor, Roger Lovatt, told me that power is the ability to make people do something that they don’t want to do, and that’s really the question here. Does OpenAI have the ability to get consumers, developers and enterprises to use its systems more than anybody else, regardless of what the system itself actually does? Microsoft, Apple and Facebook had that. So does Amazon – this is a real flywheel. <\/p>\n","protected":false},"excerpt":{"rendered":"As we all know, OpenAI has been running around trying to join the club, claiming a few months…\n","protected":false},"author":2,"featured_media":486770,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[62,276,277,49,48,61],"class_list":{"0":"post-486769","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-ca","12":"tag-canada","13":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/486769","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/comments?post=486769"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/486769\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media\/486770"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media?parent=486769"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/categories?post=486769"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/tags?post=486769"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}