Intel moves into GPU production with new chief architect hire to challenge Nvidia's AI dominance

Intel is moving into GPUs and has hired a chief architect, CEO Lip-Bu Tan says

Lip-Bu Tan, chief executive officer of Intel Corp., departs following a meeting at the White House in Washington, DC, US, on Monday, Aug. 11, 2025. Intel CEO Lip-Bu Tan said Tuesday that the chipmaker has appointed a new chief architect to build out graphics processing units, or GPUs. The chips, made by the likes of Nvidia and Advanced Micro Devices, power large language models and have skyrocketed in demand as companies race to build out artificial intelligence infrastructure and data centers. Tan told the audience at the Cisco AI Summit that it took "some persuasion" to convince the new executive to join the chipmaker. Tan did not name the new hire. Intel's stock has rallied over the last year as investors gain optimism about the company's foundry business, but the company is primarily making chips for itself. Over the past few years, the embattled American chipmaker has fallen behind major semiconductor players that have seen a boost from the AI data center buildout. Earlier this month, production snags and supply troubles overshadowed the chipmaker's better-than-expected quarterly results. Investors had also been hoping for more clarity on an anchor customer for its foundry segment.

Intel CEO Tan announces GPUs to rival Nvidia

At the Cisco AI Summit on Tuesday, Intel CEO Lip-Bu Tan announced that the company will begin producing graphics processing units, or GPUs, which rival Nvidia currently dominates. GPUs serve as specialized processors, differing from the central processing units, or CPUs, that Intel has traditionally manufactured. These chips support applications such as gaming and the training of artificial intelligence models. TechCrunch contacted Intel for additional details on the announcement. The GPU project falls under the oversight of Kevork Kechichian, executive vice president and general manager of Intel's data center group. Reuters reported on Kechichian's role in this effort. Intel hired Kechichian in September as part of multiple new hires focused on engineering expertise. Intel brought on Eric Demers in January to contribute to the GPU initiative. Demers spent more than 13 years at Qualcomm, where he most recently held the position of senior vice president of engineering. The effort remains in relatively early stages. Tan stated that Intel plans to develop its strategy around customer demands and needs. Nvidia did not invent the GPU, yet this type of chip has contributed substantially to the company's achievements. Nvidia's GPUs designed for AI systems stand out for their advanced capabilities and widespread adoption, securing the chipmaker a commanding lead in the market.

Intel Shows No Signs of Leaving the GPU Business, as CEO Lip-Bu Tan Reiterates Commitment, Saying Future Lineups Will Be Built Internally

Intel's GPU business has always been a mystery, not just on the consumer front, but also in the datacenter segment, but it appears Tan has a plan in mind. Intel has been in a struggling position ever since Lip-Bu Tan took over, given that the company still hasn't defined a clear roadmap for where it is heading with AI. When you look at Team Blue's accelerator plans, the only thing we know about is the 'inference-focused' Crescent Island, since progress around Jaguar Shores is still uncertain. On the consumer front, well, the company launched Battlemage SKUs at the end of 2024, and Pro variants in H2 2025, but the launches were an indicator of Intel's potential, not its commitment. Speaking at the Cisco AI summit, Intel's CEO Lip-Bu Tan is still working on GPUs internally, and when asked about the involvement of the foundry division to produce upcoming lineups "at scale", Tan affirmed this angle. He also revealed that the company's latest executive hire from Qualcomm, Eric Demers, is part of a broader strategy to return to the GPU segment, indicating that Team Blue isn't leaving the consumer/AI markets behind.' Question: So Intel is gonna build GPUs, CPUs, have their own foundries to build them, and going to have foundries to build at scale for other manufacturers, and will make sure you partner with other GPU providers. Intel's CEO: That's right. I just hired the chief GPU architect, and he's ⁠very good. I'm very delighted he joined me, and it takes some persuasion At the Intel Tech Tour, Intel announced that it is adopting an annual product cadence, disclosing development on Crescent Island and the next-gen Jaguar Shores lineup, indicating that the company is looking towards the AI accelerator TAM. However, given how tough the competition has become in a segment dominated by NVIDIA/AMD and Intel, Intel's re-entry would need to nail execution across TCOs, scale, and infrastructure offerings. Yet again, Intel's next step should be to develop a well-defined roadmap, whether for gamers or for enterprise. On the consumer front, we saw Intel achieve massive success with the launch of Xe3 'Celestial' iGPUs with Panther Lake, with the Arc B390 showing performance levels rivaling AMD's Strix Halo in many cases. As far as the discrete GPU segment is concerned, we are still expecting a Battlemage model, the Arc B770, which could see an unveiling at this year's Computex. Overall, it is safe to say that Intel's GPU business is still alive, but whether it can sustain itself is something future releases will decide.

With Investments from Nvidia, Intel to Start Making GPUs for the AI Market

This news comes at a time when Nvidia is facing its own challenges around GPUs that OpenAI and others claim are low on inference capabilities Last September, the Trump administration threw a lifeline to Intel by acquiring a 10% stake in the company that was fast losing ground. Others like Nvidia joined hands by making a significant $5 billion investment late last year to co-develop products that meet AI datacentre demand. Now, the company plans to start making the GPUs in a market dominated by the very same Nvidia. Nvidia had pumped in all that money for a roughly 4% stake in Intel as it saw an opportunity to access the legacy chipmaker's manufacturing capability while also hoping to enhance its offering in the AI-driven PCs and datacentre solutions. Now Intel CEO Lip-Bu Tan has confirmed that his company would soon start producing GPUs. The announcement came at the CISCO AI Summit where he highlighted the importance of being future ready given the fast pace of AI advancements. He expressed satisfaction that customers continued to engage with Intel but underscored that his company was "also focused on the 14A node, that will be very important to us. This is the most advanced node. We're going to be in risk production in 2028." GPUs, which are used for gaming and training AI models, are considerably more specialized processors than the traditional CPUs Intel has made all these years. A report by Reuters said the project would be overseen by EVP and general manager Kevork Kechichian, who is part of several other top engineers that the company hired in September. Tan revealed that the initiative was kickstarted only recently and Intel would develop its strategy around customer needs and demands. He noted that the AI wave was sucking up a lot of memory and if there is anything that could potentially slow down the AI march, it was going to be memory. This is the reason Intel needs to overcome any production challenges of processing capacity as compute is now very important. "We also want to find new ways of liquid cooling because air cooling is not cutting it anymore. Data centres need a lot of power - so the question becomes, how can we scale?" he asked. Tan also confirmed that Intel would be looking at different materials in its process of building CPUs and GPUs required for AI expansion. There needs to be a different outlook now with lots of innovation. But it is important to remain curious and think of the outcomes we want and then go back to the foundation, the Intel chief said during a session at the AI Summit. It was at the CES 2026 that we first heard of Intel's GPU dreams. At the annual even, Lip-Bu Tan highlighted out the company's Panther Lake Core Series 3 processors would be ideal for the gaming industry. Panther Lake chips are the first built by the company on their 18A manufacturing process that went into production sometime in 2025. Now Tan is talking about the 14A node, given that Intel had originally committed to consolidating and focusing on its core business when he had just taken over as CEO last March. We can only wonder whether this shift has the tacit approval of Nvidia, given that the company itself is facing a challenge around its GPUs lacking the necessary inference power for AI expansion.