Intel

Intel Corporation is an American multinational semiconductor company founded in 1968 by Robert Noyce and Gordon Moore, and the world's largest manufacturer of x86 microprocessors. In the artificial-intelligence era, Intel supplies AI accelerators through its Gaudi product line, embeds AI acceleration into its Xeon server CPUs, and operates the advanced packaging and foundry capabilities on which much of the global AI hardware supply chain depends.^[1]

Intel is the second-largest merchant chipmaker by revenue after years of x86 dominance, but it lost its early lead in AI accelerators to NVIDIA and spent the first half of the 2020s restructuring around three priorities: catching up in data-centre AI compute, restoring manufacturing leadership through its foundry business, and defending its enormous CPU installed base as AI workloads spread to every server.

History

Intel was founded in 1968 and created the first commercial microprocessor in 1971. Through the 1980s and 1990s its "Intel Inside" partnership with Microsoft defined the personal computer, and by the 2000s Intel was the world's most valuable chip company. The company struggled to transition to mobile computing and fell behind Taiwan Semiconductor Manufacturing Company (TSMC) in manufacturing process technology during the 2010s, with repeated delays in moving from 14-nanometre to 10-nanometre and 7-nanometre nodes.

Pat Gelsinger, chief executive from 2021 to late 2024, launched an "IDM 2.0" strategy that reopened Intel's factories to external customers and committed tens of billions of dollars to new manufacturing and packaging sites, including one of the largest-ever foreign investments in Malaysia. Intel's AI-specific work centred on Habana Labs, an Israeli AI-chip startup acquired in 2019, whose products were rebranded as the Gaudi accelerator family. After Gelsinger's departure, Lip-Bu Tan was appointed chief executive in March 2025 to continue the turnaround, and Intel reported progress in 2025-2026 toward volume production on its Intel 18A process node and High-NA EUV lithography, becoming the first company to produce high-volume logic with the next-generation lithography tooling.^[2]

Key Technologies

Intel's dedicated AI accelerators are the Gaudi series. Gaudi 2 launched in 2022, and Gaudi 3 followed in late 2024, built on TSMC's 5-nanometre process with two compute dies containing eight matrix-multiplication engines and 64 tensor processor cores, 128 GB of HBM2e memory, 3.7 TB/s of memory bandwidth and about 1.8 petaflops of FP8/BF16 compute. Gaudi systems scale out over standard Ethernet rather than proprietary fabrics, and Intel markets them on price-performance, claiming around two times the performance per dollar of NVIDIA's H100 on common open-model inference workloads.^[3]

Alongside Gaudi, Intel's Xeon server processors increasingly include built-in AI acceleration such as the AMX matrix engines in Xeon 6, letting ordinary servers run inference without add-on accelerators, and Intel's Arc GPUs target client and inference workloads. The company's foundry arm produces chips for other designers using its own process nodes, with Intel 18A and High-NA EUV positioned as its comeback technologies, while its advanced packaging portfolio — including Foveros 3D stacking and EMIB bridges — is a critical input for AI accelerators of every brand.^[2][4]

Applications and Impact

Gaudi accelerators are deployed mainly for cost-sensitive large language model training and inference, with availability through IBM Cloud and other providers, and Intel has published reference customers in enterprise, telecom and government segments. Xeon-powered servers remain the default compute substrate of the world's data centres, meaning most AI software runs on Intel CPUs even when the heaviest workloads use other vendors' accelerators. Intel's packaging and foundry services also carry strategic weight: governments in the United States, Europe and Asia have supported Intel plants as part of efforts to diversify semiconductor supply chains away from East Asia.^[1][4]

Intel's market position in AI is that of a broad supplier rather than a leader in any single workload. Its share of discrete AI accelerators is small relative to NVIDIA and AMD, and its turnaround has been complicated by large losses in manufacturing, but its scale in CPUs, packaging and global factories keeps it central to the AI hardware ecosystem.^[1]

>See Also

• AMD
• NVIDIA
• NVIDIA Blackwell
• Tensor Processing Unit
• AI Data Centres Malaysia