Report Says Google Plans Intel-Made TPUs for 2028, Driving a Jump in Intel Shares

Intel shares rose sharply on Monday after a Reuters report, citing The Information, said Google has placed an order with Intel to manufacture more than three million tensor processing units in 2028. The move would represent additional volume for Intel’s contract manufacturing business, an area it has been working to scale as the AI chip boom stresses semiconductor supply chains.

The reported order is focused on Google’s specialized AI accelerators. Tensor processing units, or TPUs, are purpose-built chips designed to speed up machine learning workloads, including training and running large AI models. Google Cloud documentation shows that its TPU systems are built in large “pod” architectures, where many chips are interconnected to deliver high-throughput computation for model work. The implication of a TPU foundry order is that Google is seeking manufacturing capacity beyond its usual chip in-house pathway.

According to the Reuters report, Intel’s shares climbed more than 9% early on Monday, extending a gain that would put the stock up nearly 169% for the year at the time of publication. Reuters said Intel declined to comment on the report, and said Alphabet and Nvidia did not immediately respond to requests for comment. It also noted that Reuters could not independently verify the claim, underscoring that the arrangement remains unconfirmed by the companies involved.

The report also suggested broader interest in Intel’s ability to manufacture advanced AI chips. It said Nvidia is evaluating whether Intel’s technology can be used to make a processor that combines four graphics chips into a single unit, though it added that Nvidia had not placed an order yet. In the same framing, the report said the AI-driven demand surge has left Taiwan Semiconductor Manufacturing struggling to deliver enough supply, prompting some major AI chip designers to look for diversification options.

Intel has positioned its foundry strategy as a “systems foundry” model, bundling process technology with advanced packaging and test capabilities. The company’s published materials say Intel Foundry is designed to support high-yield production of “systems-of-chips” designs and to reduce choke points across the silicon supply chain. In that context, a TPU manufacturing relationship with a hyperscaler like Google would be a high-profile validation of Intel’s ability to convert foundry capacity into real output for demanding AI workloads.

Intel’s process roadmap has included “Intel 14A” as part of its later-decade leading-edge plan. In an earlier company release, Intel said it unveiled an expanded process roadmap and highlighted Intel 14A alongside packaging and assembly-and-test upgrades tied to the company’s foundry push. While Monday’s report centers on a 2028 timeline, it arrives during a period when Intel is trying to prove that customers will commit to volume manufacturing through the hard phases of scaling new process nodes.

For markets, the key question is whether the report becomes formalized. If Google and Intel confirm the order size, chip specifications, and production milestones, it could shift expectations for Intel Foundry and potentially intensify pressure on the incumbent manufacturing concentration. What to watch next includes any Intel foundry updates around capacity and process readiness, and whether other major AI hardware suppliers announce additional “second-source” manufacturing discussions tied to TPUs or adjacent accelerators.