Intel’s GPU ambitions have often felt like a puzzle. The company has shipped products, teased next steps, and talked up big opportunities, yet the overall direction—especially in AI and the datacenter—has been hard to pin down. Now, Intel CEO Lip-Bu Tan is signaling that the GPU business is not only still active, but part of a broader strategy that ties together chips, software, and Intel’s expanding role as a manufacturer.
A key clue comes from Tan’s recent comments at the Cisco AI summit, where he confirmed Intel is continuing to work on GPUs internally and intends to use its foundry capabilities to produce future lineups at scale. That matters because “at scale” is the line between a promising technology demo and a real competitor in the market. Intel’s goal isn’t just to design GPUs and accelerators—it’s to build them reliably in volume while also operating foundries that can manufacture for other companies.
Tan also addressed Intel’s recent effort to bring in senior GPU talent from Qualcomm. The company hired veteran executive Eric Demers, and Tan made it clear this wasn’t a random leadership move—it’s part of a deliberate push to strengthen Intel’s GPU strategy. In a direct exchange, Tan confirmed Intel’s intent to build GPUs and CPUs, manufacture them through its own foundries, scale manufacturing for external customers, and still partner with other GPU providers when it makes sense. He emphasized he had just hired a chief GPU architect and described the hire as a significant win for Intel.
This all lands at an important moment for Intel. Since Tan took over, the company has been under pressure to define a clearer AI roadmap and prove it can compete in a market where NVIDIA and AMD set an aggressive pace. On Intel’s accelerator side, the most visible effort has been Crescent Island, described as inference-focused, while the status of Jaguar Shores has been less certain. Intel has also spoken about moving faster overall: during the Intel Tech Tour, the company discussed adopting an annual product cadence and referenced ongoing development around Crescent Island and the next-generation Jaguar Shores lineup. The message is that Intel wants a bigger slice of the AI accelerator total addressable market—but getting there will require strong execution across performance-per-dollar, platform readiness, and infrastructure support.
On the consumer graphics side, Intel’s recent actions show both progress and unanswered questions. The company launched Battlemage discrete GPU models toward the end of 2024, with Pro variants expected in the second half of 2025. Those releases demonstrated Intel can deliver competitive GPU products, but they also left observers wondering how deep Intel’s long-term commitment is, especially in the face of intense competition and rapid product cycles.
At the same time, Intel has had clear wins in integrated graphics. The company has highlighted strong results from its Xe3 “Celestial” iGPU work tied to Panther Lake, and the Arc B390 has been discussed as delivering performance that can rival AMD’s Strix Halo in a number of scenarios. That kind of momentum in iGPUs helps Intel in laptops and mainstream PCs, where integrated graphics performance increasingly shapes buyer decisions.
For discrete GPUs, attention is now on what comes next. One of the most anticipated products is the Arc B770, a Battlemage model that could potentially show up around Computex. If it appears, it will be another important signal: not just about performance, but about whether Intel is prepared to keep investing in a segment that demands constant iteration, strong driver support, and clear positioning for gamers and creators.
Put it all together, and Intel’s GPU business looks very much alive. The bigger question is sustainability. Intel’s next few releases—across AI accelerators, integrated graphics, and discrete Arc GPUs—will determine whether this is a temporary push or the start of a stable, long-term comeback built on a clear roadmap, consistent execution, and the ability to deliver at scale.
