Speed running AI companies on who become the leader (not actually)
The AI race is something people talk about like it’s just a tech competition but if you really sit down and think about it it’s way more than that. It’s about who controls the future of how humans think, work, and even make decisions. And when you actually look at every single company involved in this race properly you start to realize something interesting which is that every single one of them has a very specific and very different bet they are placing on the future. So let’s actually do a speed run on each of them and understand who can actually become the leader and why.
OpenAI
OpenAI was one of the first. They came in early, they built the product that made the whole world stop and pay attention and that is genuinely impressive. By early 2026 they have around 900 million weekly active users and a revenue run rate near $20 billion and they raised $110 billion at a valuation of around $840 billion. On paper that looks like a dominant company. But here is where the real problem starts when you look at the actual numbers carefully.
Out of those 900 million users only around 40 to 50 million actually pay to use ChatGPT. So you are sitting on one of the biggest user bases in the world and the conversion from free to paid is genuinely poor when you compare the ratio. And the bigger issue is what they are doing with all the money they are raising. The capital they bring in is not really going into building OpenAI as a long term self sustaining company, it’s being used almost like a runway to build more compute, more power, more infrastructure just to keep the model running and improving. So it becomes a real question that if the cost of operating is always chasing the revenue and the free users are 20 times more than the paying ones then what is the actual ceiling here and when does the money stop being enough to sustain that kind of compute hunger.
The first mover advantage is real but it’s not permanent. And right now OpenAI’s biggest challenge is not building better AI it’s figuring out how to make a business that actually works at the scale they are operating at.
Google is honestly the most interesting case in this entire race and not enough people talk about it the right way. Google has one of the biggest user bases in the entire world and the products they have built over the decades like Chrome, YouTube, Gmail, Google Maps, Android are products that are literally impossible for any other company to replicate from scratch. The kind of infrastructure and user trust that goes into something like Google Maps or YouTube is not something you build in five years or even ten years. So when Google decides to put AI into all of those products they are not starting from zero they are plugging AI into something that billions of people already wake up every day and use without even thinking about it.
That gives Google a user adoption speed that no other company in this race can match. OpenAI has to convince you to download a new app and sign up and then convince you to pay. Google just has to update the product you are already using. Gemini already has around 750 million monthly users and that number makes complete sense because Google is not really selling AI separately it’s making AI the new version of things people never stopped using.
But here is the real risk for Google and it’s a big one. When you have that kind of user base the cost of a wrong implementation is not just bad press it’s catastrophic. We already saw what happened in the early days of Bard where one mistake in a demo wiped billions off their market value. If Google rolls out AI in a way that feels invasive, broken, or wrong inside Search or Gmail or Maps the backlash would be enormous because people don’t just use Google products casually they depend on them. So Google’s biggest advantage and biggest risk are the exact same thing which is the scale of what they already have.
AWS
AWS is doing something completely different from everyone else in this race and their approach is maybe the most quietly powerful one of all. They are not trying to build the most famous AI model. They are not trying to win a user base. They are building the roads and the electricity that every other AI company needs to operate on. AWS reported $128.7 billion in full year 2025 revenue and their Bedrock service hosts models from OpenAI, Anthropic, Google, Qwen and many more all in one place for enterprise clients to deploy.
The whole tech infrastructure of the world is already built on AWS to a very significant degree. So if every company in every industry starts deploying AI models into their products and their workflows the most natural place for them to do that is the cloud infrastructure they are already using. And AWS is already there waiting. They have signed gigawatt scale data center deals going into 2026 and 2027, they have their own custom chips in Trainium and Inferentia, and they have secured some of the biggest enterprise clients in the world.
The thing about AWS’s position is that it almost doesn’t matter who wins the model race between OpenAI and Google and Anthropic because all of them are running on cloud infrastructure and a significant portion of that is AWS. It’s like owning the land that every shop in the city is built on. The shops compete with each other but the landlord profits from all of them. If implemented right AWS is not just a player in the AI race it might be the entity that makes the most money from the race itself without ever having to win it in the traditional sense.
Anthropic
Anthropic is the most morally interesting company in this entire space. Their CEO has publicly stated that he will not allow their AI models to be used by governments to kill people, which means they have essentially walked away from defense contracts entirely. And in a world where military and government spending on AI is going to be enormous that is a very expensive principle to hold. They raised $30 billion at a $380 billion valuation in February 2026 and their annualized revenue is around $14 billion growing roughly 10 times each year. They have over 500 enterprise customers including 8 of the Fortune 10 paying very serious money.
The market where Anthropic can genuinely win is the enterprise workspace market. Their Claude models are particularly strong in writing and coding which is exactly what knowledge workers, developers, lawyers, analysts, and consultants need every single day. Claude Code alone is already powering around 4% of public GitHub commits which is a meaningful number. And the trust factor that comes from their safety reputation is genuinely valuable to banks, hospitals, legal firms and other regulated industries that cannot afford an AI that hallucinates or does something unpredictable with sensitive information.
But the limitations are also real. Anthropic has never built an image model. They are not in the consumer market in any serious way. And by refusing the defense and government surveillance market they are leaving behind what might become one of the largest AI spending categories in the world. So the question with Anthropic is whether being the safest and most trustworthy enterprise AI is enough to compete at the very top level or whether the companies that take on more risk and more markets will eventually just outscale them.
Chinese AI Companies
The Chinese AI situation is something that most Western conversations about the AI race treat as a side note and that is a mistake because the numbers are actually pretty significant. Chinese AI models went from roughly 1% of global AI workloads in 2024 to around 30% by end of 2025. Alibaba’s open source Qwen family has 700 million downloads globally. Baidu’s Ernie has 200 million monthly users. These are not small numbers.
And here is the blind spot that almost nobody is talking about seriously. If AI eventually becomes as basic and essential as water or electricity, something that just runs in the background of everything, there is a very large part of the world where people still do not even know what AI is or that it exists as a usable tool. In those regions, in parts of Asia, Africa, Latin America, the model that wins is probably going to be the one that is cheapest and easiest to access. And Chinese open source models that run on modest hardware and cost very little to deploy could fill that gap very fast.
The problem is that the data question becomes a massive political issue. All the user data from people using these models would flow back to Chinese companies operating under Chinese intelligence laws. And that is not just a privacy concern it’s a geopolitical one. It could create real tension between countries at a government level. So the mass adoption of Chinese AI models in non-Western regions might be technically very possible but politically it might become the most contested thing in global technology in the next few years.
NVIDIA
NVIDIA is the one that sits underneath everything else and that’s exactly what makes it the most fascinating and also the most volatile position in this entire race. As of May 2026 their market cap is around $5 trillion which makes them one of the most valuable companies in the history of business. Every single company we just talked about, OpenAI, Google, AWS, Anthropic, Chinese AI companies, they all depend on NVIDIA GPUs to train and run their models. NVIDIA is not just a player in the AI race it’s the ground the entire race is being run on.
But there is a decision that NVIDIA faces that could go very wrong in multiple directions. The U.S. has export controls that limit what chips NVIDIA can sell to China. And we have already seen what Chinese companies can do when pushed, DeepSeek showed the world that you can build a very capable AI model at a fraction of the cost people assumed. Now imagine if NVIDIA, for financial reasons or under political pressure or any other reason, decides to increase chip supply to Chinese companies. The capability those companies could unlock with better hardware is not a small thing. It would fundamentally shift the balance of the AI race in a way that U.S. policy is specifically trying to prevent.
And then there is the bigger picture financial risk with NVIDIA itself. Some analysts are already comparing the current run up in their stock and valuation to the dot com bubble or even the 2008 financial situation in terms of how disconnected the price can get from the underlying reality when euphoria takes over a market. If AI compute demand ever slows down, if AMD or a Chinese chip company closes the gap significantly, if a supply crunch hits or a new architecture fails, the correction in NVIDIA’s valuation could be genuinely historic. The same scale that makes it the greatest company in the world right now is exactly what would make its fall the most diabolical business collapse in recent memory.
So who actually wins
Now by going through all of these you can see a pattern which is that every single company in this race is one major decision away from either breaking ahead or breaking down. OpenAI has the brand but not the business model to match it yet. Google has the scale but one bad implementation could damage something irreplaceable. AWS has the infrastructure but no consumer identity. Anthropic has the trust but limited scope. Chinese AI companies have the volume but a political ceiling. And NVIDIA has everything underneath it but is sitting on a position that could either be the greatest business story ever told or the biggest single point of failure in the history of technology.
The AI race is not something any one company is going to win cleanly, it is too big and too distributed and too geopolitically tangled for that kind of simple ending. What is actually happening is that multiple winners are going to emerge in different categories and different regions and different use cases at the same time. The consumer AI winner might look completely different from the enterprise AI winner which might look completely different from the infrastructure winner. What you should not do is look at the biggest number, the highest valuation, the most users, and assume that equals dominance because in every single case we just walked through the biggest number comes with the biggest risk sitting right next to it. > No clean winner, no guaranteed outcome, no safe bet but still a race where the decisions being made right now are going to shape what the world’s most powerful technology looks like for the next fifty years.
Sources: