Google and Anthropic dent OpenAI’s dominance in the AI race

OpenAI’s lead wobbles amid fierce competition, soaring costs and a fight for talent.

OpenAI’s leadership in the AI race is in doubt. The startup faces its toughest pressure since ChatGPT’s launch due to fierce competition, rising costs, technical problems and a battle to retain talent, the FT reports.

In November Google unveiled its flagship Gemini 3 model, which is «considered the best in the world». Later the company released a new version of its image-generation tool — Nano Banana Pro.

Gemini 3 is becoming an increasingly sought-after AI solution and is already closing in on ChatGPT in popularity.

«The current world is quite different from what it was two years ago, when OpenAI was ahead of everyone», said Hugging Face co-founder and chief science officer Thomas Wolf.

Even before the launch of Gemini 3, Sam Altman told employees the company «needed to stay focused amid short-term competitive pressure». He stressed the situation would be «a little tense».

Following recent events, Alphabet’s shares jumped and its market capitalisation neared $4trn for the first time. Wall Street believes the company can combine its dominant position in internet search, cloud infrastructure and smartphones to deliver new AI capabilities to billions of existing users.

«The ability to engage with consumers, clients and companies at such scale is really what we can do thanks to our comprehensive approach», said Google AI architect and DeepMind CTO Koray Kavukcuoglu.

He also cited the use of in-house chips to train neural networks as another advantage.

«The pressure has definitely shifted to Sam Altman and his ability to monetise and keep everything afloat», said MoffettNathanson co-founder and analyst Michael Nathanson.

Salesforce CEO Marc Benioff said he had used ChatGPT every day for three years, but after spending two hours with Gemini 3 he «decided not to go back». In his words, Google has made a «crazy leap», and «the world has changed again».

OpenAI welcomes competition — but that is not quite accurate

OpenAI’s director of research said the company welcomes competition because it pushes the entire ecosystem forward.

«Our models continue to set standards in performance, reliability and practical usefulness. We will release even more advanced solutions».

However, employees feel pressure to compete on several fronts with wealthy rivals able to spend tens of billions of dollars on building AI.

Some experts believe OpenAI has overreached in its bid to scale at any cost. Over the past year it has rolled out numerous releases, from automated programming tools to the viral Sora app.

The firm has promised to spend $1.4trn over the next eight years on compute, striking major deals with Nvidia, Oracle, AMD and Broadcom.

«The company is stretching itself too thin. It is impossible to do everything well», the FT quoted a Silicon Valley venture-firm partner as saying.

Sarah Myers, co-director of the non-profit AI Now Institute, called such an approach «too risky» for any firm.

Another strong competitor

Anthropic — another fast-growing AI startup — has built a large enterprise business.

Its chatbot Claude was overshadowed by ChatGPT, but the company’s longstanding focus on safety has produced a more reliable tool for corporate clients. The firm’s programming tools are considered best in class.

OpenAI still leads by weekly active users, at 800m, but people spend more time in Gemini chat.

Author and Stanford AI institute professor Erik Brynjolfsson noted it is too early to write off OpenAI, as its extensive set of new apps is a good way to find sources of revenue.

In November, leaked documents about Altman’s company’s financial position appeared online. They suggest the startup may still be loss-making.