In an era of anxiety about unchecked corporate power, artificial intelligence can seem like the most terrifying example of all. Already valued in the trillions of dollars, the industry has unparalleled influence over our collective futures, and the government’s doing nothing to rein it in. If that’s not akin to monopoly power, what is? Yet the real story of the most consequential technology of our time is strikingly different from what it seems. Instead of consolidating, as so many other industries have done, the leading edge of A.I. has become fiercely competitive. The result has been a staggering pace of innovation, significant reductions in costs and an expanding array of choices for consumers and businesses alike.
Five years ago, worries about sparse digital competition were well founded. A handful of giants — Amazon, Apple, Google, Meta and Microsoft — dominated the tech economy. Most major product categories had only two or three serious competitors, such as search (Google and Microsoft’s Bing) and mobile operating systems (Apple’s iOS and Google’s Android). When new markets like cloud computing emerged, incumbents quickly took control.
These leads were large. Google handled roughly 90 percent of search queries. And they were stable. Facebook users could not take their social graphs to a rival platform, and I am not sure how I would pry my digital life out of Apple’s ecosystem if something better came along. Certain platforms became near-mandatory gateways. Many businesses attempt e-commerce at their peril unless they go through Amazon.
I thought if anything would lock in those advantages, it would be A.I. I could not have been more wrong.
Consider the widely followed Arena leaderboard, where chatbots compete in blind, head-to-head tests. The top-ranked lab is Anthropic, a company founded just five years ago. OpenAI, which is third, is only about a decade old. A year ago, a dark-horse entrant from China pushed into contention with Google with vastly fewer resources. Some observers concluded that large companies could never move fast enough to keep up. Google, which published research in 2017 that almost everyone since has built on, responded by behaving like a start-up again. Its co-founder Sergey Brin rolled up his sleeves and got back in on the action. That made not just the company — now in second place — but the whole field more competitive.
There are no lazy monopolists in the A.I. space coasting on past advantages. Over the past year, the top spot on the Arena leaderboard has moved among those three companies, with strong performances from newer arrivals such as the Chinese company DeepSeek and the French firm Mistral — many of which require far less capital than earlier generations of A.I. companies.
Moreover, no single company dominates across A.I. areas. Anthropic currently leads in text and coding, OpenAI in text-to-image generation, xAI (founded three years ago) in image to video and Google in search-integrated A.I.
A few weeks ago, much of my feed on X was gushing about Claude Code as a research assistant and coding agent. Many of those users now seem to have shifted to OpenAI’s competing product, Codex. Switching based on features and pricing is common. Many people and businesses now use multiple models at once, a practice known as multihoming.
When you ask a question through a service like Perplexity, it may route your query to OpenAI’s, Anthropic’s, Meta’s or Mistral’s models, depending on what it expects will perform best. It is increasingly common for developers to use routers to whatever A.I. is cheapest or fastest. Business users, meanwhile, are far less locked into A.I. systems than they are into cloud services or major software platforms.
This ease of switching has forced companies to pass the gains from innovation on to users. Free tiers now offer capabilities that recently would have seemed almost unimaginable. OpenAI pioneered a $20-per-month subscription three years ago, a price point many competitors matched. That price has not changed, even as features and performance have improved substantially.
One recent analysis found that “GPT-4-equivalent performance now costs $0.40/million tokens versus $20 in late 2022.” That is the equivalent of a 70 percent annual deflation rate — remarkable by any standard, especially in a time when affordability has become a dominant public concern.
And this is only the foundational model layer. On top of it sits a sprawling ecosystem of consumer applications, enterprise tools, device integrations and start-ups aiming to serve niches as specific as gyms and hair salons.
Users aren’t the only ones switching. The people who work at these companies move from one to another, a sharp contrast to work in Silicon Valley during the era of do-not-poach agreements. Dario Amodei, the chief executive of Anthropic, used to work at OpenAI. Leaders from OpenAI, Meta and elsewhere have gone on to raise large sums for rival or complementary ventures.
This churn has helped prevent any single technological paradigm from taking control. Enormous sums of venture capital have flowed readily to alternative approaches — like world models that aim to reason about reality more directly than large language models — and incumbents cannot afford to ignore them.
For a while, Nvidia was the provider of the most desired chips, especially for the more processing-power-intensive model training runs. Late last year, however, Google’s Gemini 3 model vaulted to the top of the leaderboards by relying on a new custom-designed chip. When Anthropic overtook Google for the No. 1 spot, it did so using chips from several companies. Older companies like AMD are re-emerging as formidable designers, as are lean new A.I.-first entrants like Cerebras that are specializing in the inference the A.I. systems use to answer specific queries.
The A.I. sector may actually be too competitive, at least in the short run. Major companies lose money on each new A.I. customer because they expect today’s market share to translate into future profits. Sky-high valuations suggest investors believe them. But all the talk that we might be witnessing a market bubble proves the matter is far from settled.
Two forces may ultimately keep A.I. from financial success, even if it all works out technologically. One is the rise of open-weight models — which allow users to run and even customize the underlying systems, from companies like Meta, DeepSeek and Mistral — that cost much less to develop. They can’t perform the most advanced functions but are more than good enough for most uses and vastly better than state-of-the-art models from just a year ago. They are likely to continue placing downward pressure on prices across the board.
Another is the similarity among products. When consumers don’t see much difference among options, they default to whichever is least expensive. This pattern, which economists refer to as commoditization, is why companies are racing to differentiate their products. One way they’re doing that is by personalizing their products, which can make them more satisfying to use but also harder to leave.
At the moment, it costs hundreds of millions if not billions of dollars to enter this market, a barrier that limits the number of contenders. The success of more agile approaches may change that. If it becomes easier for new companies to jump in, then we can expect competition to go way up and profit margins to be even more elusive.
Competition is great at getting people what they want — for better or worse. If people want help with recipes, book recommendations or email drafts, a competitive market will deliver. If they want deepfakes, A.I.-generated spam and hyperaddictive misinformation, competition will deliver those, too. Neither competition nor monopoly is any guarantee against the rapid dislocation so many are worried about. Regulation may be necessary in some areas, but if so, it won’t be to break a monopoly.
You can’t enjoy the benefits of intense competition without accepting that there will be winners — and many losers. For now, policymakers and the public should recognize just how unusual this moment is. A.I. competition is delivering rapid innovation, falling prices and real choice at a pace few expected. Over time, policy will need to adapt to protect users, preserve competition and spread the gains. For the moment, the market is doing that far more than anyone would reasonably have thought possible.
Jason Furman, a contributing Opinion writer, was the chairman of the White House Council of Economic Advisers from 2013 to 2017.
The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: [email protected].
Follow the New York Times Opinion section on Facebook, Instagram, TikTok, Bluesky, WhatsApp and Threads.
The post The Race to Dominate A.I. Is Brutally Competitive. That’s Good for Everyone. appeared first on New York Times.
