Skip to main content
Saavage editorial graphic for Three AI races at once.
AI Watch

Google, OpenAI, and Anthropic: who is actually winning?

The AI race is not one scoreboard. OpenAI, Anthropic, and Google are each trying to win a different layer of the stack, which is why the answer depends on what you think AI becomes next.

OpenAI still has the consumer lead, Anthropic is gaining in enterprise AI, and Google is betting hard on infrastructure. The real race is about distribution, trust, agents, and compute.

SourceRefreshing AI Latest

Subscribe to the channels

Key Points

  • OpenAI still owns the consumer front door.
  • Anthropic is winning trust with enterprise teams and agent workflows.
  • Google is betting that compute, chips, and data centers become the real choke point.

There is no single AI leaderboard anymore

The AI race is getting talked about like one company is about to win the whole thing. I do not think that is how this plays out. OpenAI, Anthropic, and Google are not even making the same bet.

OpenAI owns the consumer conversation. Anthropic is becoming the enterprise default in places where people are nervous about reliability and agent safety. Google is spending like the company that thinks compute is the real choke point. All three can be right for a while.

OpenAI still owns the consumer front door.
Saavage field notes graphic: There is no single leaderboard anymore.
There is no single leaderboard anymore

OpenAI still has the front door

OpenAI's advantage is that regular people know the product. ChatGPT is still the default verb for AI, and that kind of mindshare is almost impossible to buy after the fact. Developers, students, businesses, creators, everyone has at least tried it.

That gives OpenAI distribution. The risk is that distribution alone does not guarantee the best enterprise contracts, the cheapest compute, or the safest agent platform. Being the front door is powerful, but the whole house is getting bigger.

Saavage field notes graphic: How I read the race.
How I read the race

Anthropic is winning the trust lane

Anthropic's rise makes sense when you look at what businesses actually ask for. They want long context, predictable behavior, good coding, fewer weird surprises, and a vendor that sounds like it takes risk seriously. Claude fits that lane.

That does not make Anthropic boring. It makes it dangerous. Enterprise defaults are sticky. Once a big company builds internal workflows around Claude, it does not swap them out because another model wins a benchmark for a week.


Google is playing the long game with the expensive part

Google's bet is less glamorous, but it might age the best. Models change quickly. Apps change quickly. Data centers, chips, power contracts, and cloud capacity do not. If compute stays scarce, owning the rails matters.

The question is timing. Google can afford to wait, but AI product cycles are moving in months, not years. If the market rewards the best app first, OpenAI has the edge. If it rewards durable infrastructure, Google is still very much alive.