Intel’s Shares Rise on Reported Google TPU Order, Boosting Foundry Momentum Hopes

Intel shares jumped Monday as investors latched onto a new report that Alphabet’s Google has placed an order with Intel for more than three million Tensor Processing Units, or TPUs, for production in 2028. The report, attributed to The Information and carried in a Reuters write-up, framed the potential order as a validation of Intel Foundry’s progress and a sign that major AI chip buyers may be seeking alternatives to the current concentration of advanced manufacturing capacity.

TPUs are specialized chips designed for running machine learning workloads, particularly the training and inference phases used to build and serve AI models. Under the reported scenario, Google would be using Intel not only as a technology partner but as a manufacturing backstop for a large run of its in-house accelerators, with the timing anchored to 2028 output.

According to the Reuters account, Intel’s stock rose more than 9% in early trading on the news. The same report said Intel declined to comment, while Alphabet and Nvidia did not immediately respond to requests for comment, and that Reuters could not independently verify the claim. It also noted that the order, if confirmed, would bolster Intel’s contract chip manufacturing business as the company tries to regain ground after years in which it lost manufacturing leadership to Taiwan Semiconductor Manufacturing Company, or TSMC.

The Reuters write-up also said Nvidia was evaluating whether Intel technology could be used to make a processor that combines four graphics chips into a single unit, though it had not placed an order at the time of the report. This broader context matters because Intel is still trying to convert early “ecosystem” indicates into sustained external volume for advanced nodes and system-level assembly, where a foundry strategy lives or dies on throughput, yield, and customer commitments.

Intel has been pitching its “systems foundry” approach as a way to make advanced manufacturing and advanced packaging work as a complete pipeline rather than as disconnected pieces. In an Intel Foundry overview, the company describes the model as combining IP, electronic design automation tools, process nodes, advanced packaging, and Intel Foundry Advanced System Assembly and Test capabilities, aimed at high-yield production of “systems of chips” for customers.

For advanced AI manufacturing, Intel has also emphasized specific technology elements and packaging methods. In a foundry fact sheet, Intel describes its Intel 18A process node as pairing RibbonFET transistor architecture with PowerVia backside power delivery. The same material highlights packaging technologies such as EMIB, an embedded multi-die interconnect bridge that helps integrate components into custom multi-die systems with lower cost than monolithic designs.

Still, Monday’s move appeared to be based more on a potential anchor customer than on any new, confirmed Intel guidance. The report did not provide key commercial details that investors typically want, including which exact Intel process node or packaging configuration would be used for the TPUs, whether the number is a firm production order versus an evaluation milestone, and how much of the volume would be allocated to Intel’s internal manufacturing timeline versus partner assembly and test capacity.

What to watch next is whether Intel or Google acknowledges the discussion and whether any specifics follow, such as the manufacturing node, the assembly and packaging configuration, and the start-of-production schedule. Investors will also be looking for additional indicates from other large AI chip designers, especially any formal commitments from Nvidia or other hyperscalers, because external volume is the central requirement for Intel’s foundry turnaround to translate into durable financial improvement. Until then, the current development remains a report about a potential customer relationship, not a disclosed contract.