PHOTOGRAPH: OPEN AI/YOUTUBE
There was a time recently when AI seemed to belong to one firm. OpenAI, a San Francisco-based startup, shot to global fame in 2022 after ChatGPT’s release, dazzling the world with a chatbot that could think and write with eerie ease. Its rapid-fire upgrades widened its lead and wrong-footed even the biggest tech firms. For a brief moment, its brand became the definition of the cutting edge. That dominance is over. A serious rival is now emerging.
The biggest threat comes from Google. The search behemoth, which had seemed befuddled by the AI uprising on its doorstep, has struck back. Its newest AI model, Gemini 3, has for the first time surpassed OpenAI’s offerings across a range of industry benchmarks such as logical reasoning and advanced mathematics. Independent evaluators and corporate clients report a big leap in performance. The model can work with text, images and audio all at once; and that is not easy for others to copy. The narrative has flipped.
Google’s counter-offensive stretches far past making a stronger AI model. It is perhaps the only firm that can lay claim to controlling the entire “AI stack”. It designs its own specialised AI chips, known as TPUs. It owns vast data centres to run them. It develops the models themselves. And crucially it operates the planet’s largest digital distribution network: Search, YouTube, Gmail, Maps and the Android operating system. This allows Google to inject new AI capabilities directly into the bloodstream of the internet, putting them in front of billions of users without the friction of a standalone product. OpenAI on the other hand must rely on partners, chiefly Microsoft, for such reach.
The importance of this became obvious when Google put Gemini straight into its search engine as soon as it launched. Instead of showing only a list of links, the search engine now lets you ask questions in plain language, get answers that combine ideas, see images it creates and tackle complicated problems in one go. If people start using this new AI-powered search, it could prevent OpenAI from ever becoming the main way people find information online.
To get here, Google has undertaken a painful process of self-disruption. It has reorganised its AI divisions, merged research teams and redirected its colossal resources towards an “AI-first” strategy. The result is a company willing to cannibalise its legendary search engine, the source of the most profitable business model in history, to secure its future. The market has noticed. Google’s parent company, Alphabet, recently posted quarterly revenue of $100bn for the first time. Its cloud-computing division is growing at a record clip. Its valuation has soared past $3.5trn, overtaking Microsoft’s.
OpenAI meanwhile faces mounting headwinds. It has committed to a stonking $1.4trn in long-term computing costs, a bet that assumes its growth will continue unabated. Signs that user growth is slowing could jeopardise this financial footing. Unlike Google it controls no platform of its own, leaving it dependent on Microsoft’s infrastructure and distribution. Its aura of invincibility, a powerful soft asset in the tech world, is fading. This has prompted its own backers to hedge their bets. A $15bn investment by Microsoft and Nvidia in Anthropic, another AI lab, suggests that OpenAI is no longer the only gravitational centre in the AI universe.
The wider AI industry moreover is changing in ways that help a company like Google. As AI-generated summaries in search results trim traffic to external websites, publishers worry about their future. Yet this very dynamic underlines Google’s power. It controls the pipeline of information that AI models, including OpenAI’s, rely on for training data. Any move by Google to restrict access or attach conditions to this data would leave standalone labs in a dangerous position.
None of this means that OpenAI is doomed. It retains a formidable brand with hundreds of millions of weekly users. The AI race is a marathon, not a sprint, and technological leadership can change hands again. But for the first time since ChatGPT’s debut OpenAI is playing defence. The combination of a resurgent Google, its own immense financial commitments and a splintering competitive landscape presents the most serious threat to its position yet. The age of AI has its first great corporate rivalry. ■
