AI race is now a two-horse race

Nvidia's $5 bn lifeline to Intel will bring the venerable firm that gave Silicon Valley its name into the AI infrastructure business and could reshape the US tech landscape. The deal integrates Intel's processors with Nvidia's AI computing stack and extends to the lucrative data centre market. It stops short of being a contract-manufacturing agreement, which would have given Intel a bigger lead over rivals like AMD and Taiwan Semiconductor Manufacturing Company (TSMC), which produces Nvidia's flagship processors. Yet, the deal could morph into a tighter alliance as the two firms co-create standards for AI computing.

Intel's current woes arise from its having missed the mobile computing and AI opportunities. A bid to take over Nvidia was shot down as being too expensive. At that point, Intel was the dominant player, with AMD trailing and Nvidia churning out graphics chips. The partnership will consolidate the US chip-making industry, building greater immunity against Chinese challengers who resent Nvidia's control over AI computing. This sits well with the Trump administration, which has become the biggest shareholder in Intel. The AI race is now diverging into US and Chinese hardware stacks, with strong support from their respective governments. Access to computing has become a bargaining chip that neither economy can afford to ignore.

Stodgy US tech companies are now being retrofitted for AI. Oracle is being tapped for its cloud-computing infrastructure, and now Intel will get to make chips for data farms. In China, Huawei is powering ahead with its advanced chips, while Alibaba and Baidu pursue their own semiconductor development. Pitted against the American grandees are Chinese upstarts who are quicker to the draw. Increasingly, AI is becoming a two-horse race, with governments in the saddle. No other economy has the resources to compete. There may be some opportunity in contract manufacturing that is open to the rest of the world as the drawbridges are pulled up.