AI Chip Makers

Advanced Micro Devices

Santa Clara, United States

Advanced Micro Devices (NASDAQ: AMD) is one of the worlds largest semiconductor companies and the second-largest manufacturer of AI accelerator GPUs after Nvidia, founded in 1969 and headquartered in Santa Clara, California. Led by CEO Dr. Lisa Su, AMD manufactures the Instinct MI300X, MI325X, and the upcoming MI450X AI accelerators competing directly with Nvidia in AI training and inference. The MI450X scheduled for late release is expected to significantly close the gap with Nvidia. AMD also manufactures EPYC server CPUs, Ryzen consumer CPUs, and Radeon GPUs. The company signed a multi-year deal in February to supply Meta Platforms with GPUs for AI data centers.

Cerebras Systems

Sunnyvale, United States

Cerebras Systems is a leading AI chip company specializing in wafer-scale processors for AI training and inference, founded and headquartered in Sunnyvale, California. Founded by Andrew Feldman, Sean Lie, Gary Lauterbach, Michael James, and Jean-Philippe Fricker, Cerebras manufactures the Wafer Scale Engine (WSE-3), the worlds largest computer chip with 4 trillion transistors and 900,000 cores delivering ultra-fast inference for large language models. The company offers Cerebras Inference (the worlds fastest AI inference service), the CS-3 system, and Cerebras Condor Galaxy AI supercomputers. Cerebras filed confidentially for an IPO with Citigroup as lead underwriter and was reportedly preparing to launch the public offering in April.

Etched

San Francisco, United States

Etched is an AI chip startup specializing in transformer-specific ASIC chips designed exclusively for running transformer-based AI models, founded and headquartered in San Francisco, California. Founded by Gavin Uberti, Chris Zhu, and Robert Wachen as Harvard sophomores, Etched manufactures the Sohu chip, the worlds first transformer-specific ASIC delivering an order of magnitude faster inference speeds than Nvidia GPUs for transformer workloads at significantly lower cost. The company has raised 120 million in Series A funding from Primary Venture Partners, Two Sigma Ventures, Hummingbird, Fundomo, and other investors with strong demand from AI inference customers seeking specialized silicon for production transformer model deployment.

Groq

Mountain View, United States

Groq is an AI inference chip and platform company specializing in ultra-fast Language Processing Unit (LPU) chips, founded and headquartered in Mountain View, California. Founded by Jonathan Ross (creator of the Google TPU) and Douglas Wightman, Groq manufactures custom LPU silicon designed specifically for AI inference workloads achieving record-breaking inference speeds for popular open-source models including Llama, Mistral, and DeepSeek. The company operates GroqCloud providing API access to its LPU infrastructure, GroqRack on-premise systems for enterprises, and partnerships with Meta, Aramco, and Saudi Arabias HUMAIN. Groq has raised over 1.5 billion in funding at a 6.9 billion valuation.

NVIDIA

Santa Clara, United States

NVIDIA Corporation (NASDAQ: NVDA) is the worlds largest semiconductor company by market capitalization and the dominant supplier of AI accelerator chips, founded in 1993 and headquartered in Santa Clara, California. Led by founder and CEO Jensen Huang, Nvidia became the first company to reach 4 trillion and 5 trillion market capitalization driven by AI demand. The company manufactures the H100, H200, B100, and B200 GPUs, the GB200 NVL72 system for AI training, the Spectrum and Quantum networking switches, and the CUDA software platform powering the AI industry. Nvidia chips power virtually every major AI laboratory including OpenAI, Anthropic, Google, Meta, Microsoft, and xAIs Colossus supercomputer.

SambaNova Systems

Palo Alto, United States

SambaNova Systems is an AI chip and full-stack AI platform company, founded and headquartered in Palo Alto, California. Founded by Stanford professor Rodrigo Liang, Kunle Olukotun, and Christopher Re, SambaNova manufactures the SN40L Reconfigurable Dataflow Unit (RDU) chip designed for both AI training and inference of large language models including 1 trillion+ parameter models. The company offers SambaNova Cloud for AI inference, the DataScale system for on-premise AI deployment, and SambaNova Suite for enterprise generative AI. Customers include the US Department of Energy, Argonne National Laboratory, Saudi Aramco, and major financial services firms. SambaNova has raised over 1.1 billion in funding at a 5 billion valuation.

Tenstorrent

Toronto, Canada

Tenstorrent is an AI chip company building open-source AI processors, founded and headquartered in Toronto, Ontario, Canada. Led by CEO Jim Keller (legendary chip designer behind Apple A4/A5, AMD Zen, and Tesla FSD), Tenstorrent develops AI processors based on the open-source RISC-V architecture including the Wormhole and Blackhole chips, and the Grayskull product line. The company sells AI accelerators direct to customers, offers cloud access through Tenstorrent Cloud, and licenses its IP to other chip designers including Bosch, Hyundai, and Samsung. Tenstorrent has raised over 700 million from Jeff Bezos, Samsung, Hyundai, Fidelity, and other investors at a 2.6 billion valuation as a leading Nvidia alternative.