--- title: "Emerging chip manufacturers continue to challenge NVIDIA, with the main player this round being D-Matrix, which is invested by Microsoft" type: "News" locale: "en" url: "https://longbridge.com/en/news/289205648.md" description: "The AI chip startup D-Matrix, invested by Microsoft, claims that its \"Pirate Ship\" inference chip achieves speeds 10 times that of NVIDIA's standalone GPUs in small computing tasks, with only one-fifth of the energy consumption. The company aims to carve out a niche in the NVIDIA-dominated market, with the founder stating that the trillion-dollar market can accommodate new entrants, and there are no plans for sale at this time" datetime: "2026-06-09T14:23:35.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/289205648.md) - [en](https://longbridge.com/en/news/289205648.md) - [zh-HK](https://longbridge.com/zh-HK/news/289205648.md) --- # Emerging chip manufacturers continue to challenge NVIDIA, with the main player this round being D-Matrix, which is invested by Microsoft **Key Points** 1. D-Matrix claims that its chips can perform inference tasks at speeds up to 10 times that of NVIDIA's standalone GPUs, with energy consumption only one-fifth of the latter, but this advantage only applies to small-scale computing tasks. 2. Similar to Cerebras, D-Matrix is attempting to carve out a niche in the AI chip market dominated by NVIDIA. 3. D-Matrix co-founder and CEO Sid Shekhar stated in a CNBC interview, "A trillion-dollar market is taking shape." The competition in the AI chip sector is becoming increasingly intense, with another startup claiming its product performance surpasses that of NVIDIA, the world's highest-valued company. D-Matrix is headquartered in Silicon Valley, just three miles from NVIDIA's headquarters. The company states that for small-scale inference tasks, its chip operates at speeds 10 times that of NVIDIA's standalone graphics processors, with energy consumption only one-fifth of theirs. The new inference chip, named "Corsair," features an innovative memory architecture design, similar to that of Cerebras and Groq. The demand for computing resources from major tech giants continues to surge, clearly indicating that there is still ample space for small and medium-sized chip companies to explore niche market opportunities. Founded in 2015, Cerebras completed a significant IPO last month, raising over $5.5 billion, with a current valuation exceeding $50 billion; Groq was acquired by NVIDIA for $20 billion last December, marking the largest acquisition in the history of this AI giant. NVIDIA subsequently launched new products based on Groq technology at the GTC conference in March—language processing units. Sid Shekhar told CNBC in the interview, "A trillion-dollar market is taking shape, and I have no intention of selling the company. This market can fully accommodate another publicly listed chip company." D-Matrix was founded in 2019 and has raised approximately $500 million in total financing, with a post-investment valuation of about $2 billion. Microsoft's venture capital arm, M12, is one of the investors. This is noteworthy: Microsoft has ambitious plans for its own chip layout, developing the Maia 200 AI inference chip and collaborating with NVIDIA to create a new generation of PC processors, and just last week released its own quantum computing chip. Shekhar has not disclosed the specific customer list for the Corsair chip but revealed that they have secured procurement intentions from several leading cloud service providers, new cloud vendors, and cutting-edge AI laboratories, all of which are striving to expand their computing reserves. D-Matrix began bulk shipments to customers this month; about 90% of its customers are located in the United States, with overseas buyers concentrated in the Middle East and Southeast Asia. On June 1, 2026, during the Taipei Computer Show and NVIDIA's GTC conference, NVIDIA CEO Jensen Huang announced the RTX Spark super chip NVIDIA officially enters the PC chip market, aiming to break Intel's long-standing monopoly and adapt PC terminals to the demands of the AI era. Stacey Rasgon, a semiconductor analyst at Bernstein Research, analyzed: "Most customers will purchase D-Matrix chips to use in conjunction with NVIDIA chips, as different chips excel in different computational tasks. It seems this company has already established numerous real customer collaborations." D-Matrix's Corsair chip integrates memory and computing units highly within a single chip to achieve low power consumption and low latency inference calculations. In line with the technical routes of Groq and Cerebras, D-Matrix uses Static Random Access Memory (SRAM), which can be produced by logic wafer fabs like TSMC and integrated within the chip. In contrast, traditional GPUs rely on massive amounts of Dynamic Random Access Memory (DRAM), which must be separately packaged as high-bandwidth memory stacked around the logic chip. Currently, the global supply of DRAM capacity from Micron, Samsung, and SK Hynix is tight. Schaeffer stated: "Our products will not encounter DRAM capacity bottlenecks because the entire solution does not rely on DRAM for performance." Rick Bart, a part-time professor of electrical engineering at Stanford University, pointed out that the D-Matrix solution has a significant shortcoming: the SRAM architecture cannot support ultra-large parameter inference models. The data transmission distance of on-chip SRAM is extremely short, resulting in impressive inference speed performance, but it cannot accommodate the large models with trillions of parameters from leading companies like OpenAI and Anthropic. Bart stated: "Models with trillions of parameters cannot be accommodated on SRAM-based chip architectures; this is the biggest challenge of this technical route." Schaeffer, however, indicated that the design of Corsair is specifically for AI inference scenarios, prioritizing interactive response speed rather than supporting ultra-large language models, suitable for chatbots, voice assistants, Claude Code, OpenClaw, and other intelligent agent tools. D-Matrix cited test data from Gimlet Labs, claiming that when deployed in conjunction with NVIDIA's Blackwell GPU, Corsair's inference speed can reach 10 times that of an independent GPU, with overall usage costs reduced to one-third, and energy efficiency improved by up to 5 times. NVIDIA CEO Jensen Huang stated last week that with the flagship Vera Rubin system, NVIDIA remains the leader in the cost-effective inference market, where performance is not the only evaluation criterion. Huang mentioned at the Taipei Computer Show: "Our core advantage is our fully integrated self-research, from complete design of underlying hardware, full system simulation, to deep collaborative optimization of software and hardware, all done independently." Schaeffer introduced that D-Matrix packages four Corsair chips into a single accelerator card, which can be directly inserted into data center server rack slots, with a single card priced at tens of thousands of dollars. Schaeffer stated that this plug-and-play deployment solution is the core advantage that distinguishes D-Matrix from Cerebras and Groq; Corsair is "the highest integration density SRAM solution currently on the market," with a single server capable of accommodating up to 128GB of SRAM memory D-Matrix has also partnered with Arista, Broadcom, and Super Micro to launch the SquadRack, a complete rack-level system designed for AI data center deployment of its own chips. This chip is manufactured by TSMC using a 6-nanometer process; the next-generation product "Raptor" is planned for release next year, utilizing TSMC's 4-nanometer process. Xie Si revealed that production is expected to be arranged at TSMC's facility in Arizona, USA. Xie Si concluded: "Creating a complete computing power solution tailored for AI inference will be the industry's biggest dividend track." ### Related Stocks - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [NVDQ.US](https://longbridge.com/en/quote/NVDQ.US.md) - [NVDY.US](https://longbridge.com/en/quote/NVDY.US.md) - [07788.HK](https://longbridge.com/en/quote/07788.HK.md) - [NVD.US](https://longbridge.com/en/quote/NVD.US.md) - [NVDU.US](https://longbridge.com/en/quote/NVDU.US.md) - [NVDD.US](https://longbridge.com/en/quote/NVDD.US.md) - [NVDL.US](https://longbridge.com/en/quote/NVDL.US.md) - [NVDS.US](https://longbridge.com/en/quote/NVDS.US.md) - [NVDX.US](https://longbridge.com/en/quote/NVDX.US.md) - [07388.HK](https://longbridge.com/en/quote/07388.HK.md) - [NVDB.US](https://longbridge.com/en/quote/NVDB.US.md) - [NVDG.US](https://longbridge.com/en/quote/NVDG.US.md) - [NVDO.US](https://longbridge.com/en/quote/NVDO.US.md) - [NVDW.US](https://longbridge.com/en/quote/NVDW.US.md) - [NVYY.US](https://longbridge.com/en/quote/NVYY.US.md) - [NYYY.US](https://longbridge.com/en/quote/NYYY.US.md) - [SMH.US](https://longbridge.com/en/quote/SMH.US.md) - [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md) - [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md) - [SOXQ.US](https://longbridge.com/en/quote/SOXQ.US.md) - [XSD.US](https://longbridge.com/en/quote/XSD.US.md) - [PSI.US](https://longbridge.com/en/quote/PSI.US.md) - [FTXL.US](https://longbridge.com/en/quote/FTXL.US.md) - [MSFT.US](https://longbridge.com/en/quote/MSFT.US.md) - [TSM.US](https://longbridge.com/en/quote/TSM.US.md) - [MU.US](https://longbridge.com/en/quote/MU.US.md) - [OpenAI.NA](https://longbridge.com/en/quote/OpenAI.NA.md) - [ANET.US](https://longbridge.com/en/quote/ANET.US.md) - [AVGO.US](https://longbridge.com/en/quote/AVGO.US.md) - [SMCI.US](https://longbridge.com/en/quote/SMCI.US.md) - [INTC.US](https://longbridge.com/en/quote/INTC.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) ## Related News & Research - [Upstart chipmakers keep challenging Nvidia. This time it's Microsoft-backed D-Matrix](https://longbridge.com/en/news/289195855.md) - [ANALYSIS-Nvidia's AI PC push banks on unproven demand beyond niche users](https://longbridge.com/en/news/289042406.md) - [Nvidia’s PC chip reveal obscures a bigger catalyst for the stock](https://longbridge.com/en/news/288843922.md) - [Apple's Biggest AI Bet Yet Runs On NVIDIA, Not Its Own Chips](https://longbridge.com/en/news/289178134.md) - [Nvidia Is the 'King of Diversity,' Says 5-Star Bank of America Analyst amid Unprecedented Chip Squeeze](https://longbridge.com/en/news/289053221.md)