MatX Lands $500M to Challenge LLM Limits With New AI Chip

The competition for computational power in large language models (LLMs) is extending into more foundational and specialized chip sectors. On February 24, 2026, AI chip startup MatX, founded by a former senior engineer from Google's TPU team, announced the completion of a $500 million (approximately ¥3.445 billion) Series B funding round.
This financing round boasts an impressive roster of investors. It features strategic participation from leading semiconductor companies like Alchip and Marvell, alongside substantial investments from several top-tier venture capital firms.
Core Weapon: The MatX One Chip
MatX's confidence in this funding round is anchored in its next-generation processor currently in development: the MatX One. This chip is designed to address the dual challenge of achieving both "high throughput" and "low latency" in large model inference.
Innovative Architecture: It employs a "partitionable systolic array" design. This structure ingeniously merges the ultra-high energy efficiency of a large array with the scheduling flexibility of smaller arrays, optimizing hardware utilization.
Memory Breakthrough: The MatX One integrates an ultra-low-latency SRAM architecture with the extended context-handling capacity of High Bandwidth Memory (HBM), overcoming traditional memory bottlenecks in chip design.
Full-Scenario Adaptability: From foundational prefill tasks and high-frequency decoding to complex reinforcement learning training, the MatX One aims to deliver industry-leading performance across all workloads.
Business Outlook: Reducing LLM Operational Costs
A primary goal for model developers in today's compute market is reducing the cost per generated token. Citing MatX's official statement, IT Home reported that the company's technology has the potential to achieve throughput efficiency on par with or exceeding conventional chips, thereby significantly lowering the barrier for deploying and maintaining large-scale models.
Industry Overview: The AI Chip Arms Race Heats Up
The rise of MatX is just one part of the broader global surge in AI chip development. Recent industry activity has been intense:
SambaNova unveiled its fifth-generation RDU chip and entered a deep strategic partnership with Intel.
Positron announced its Asimov chip, claiming its performance per watt can reach five times that of NVIDIA's upcoming Rubin architecture.
Domestic Breakthrough: A research team in China recently developed a flexible AI chip with a production cost under $1. Capable of withstanding 40,000 bending cycles, it points toward new possibilities for wearable AI hardware.
Related article
Apple Smart Glasses Could Debut at WWDC27, Highlighting Privacy Protection
Bloomberg’s Mark Gurman reports that Apple’s smart glasses, codenamed N50, are slated for a WWDC27 debut in June 2027, with a retail launch expected in autumn 2027. Originally targeted for late this year and early 2027, the device’s release has been
Inside Details Exposed About Next-Gen Gemini: Strained Computing Power, Internal Teams Disagreed on Development Priorities and Resource Allocation
Reports indicate that the launch of Google’s highly anticipated next-generation Gemini model has been pushed back. Internal disagreements over development priorities and resource allocation, combined with limited computing capacity and complex approv
OpenAI Dismisses Growth Slowdown Concerns, Says Multiple Business Units Accelerating
In response to external scrutiny regarding decelerating sales growth and missed internal benchmarks, AI leader OpenAI issued a confident statement on Tuesday, April 28. The company clarified that its consumer products and enterprise services are adva
Related Special Topic Recommendations
Comments (0)
0/500

The competition for computational power in large language models (LLMs) is extending into more foundational and specialized chip sectors. On February 24, 2026, AI chip startup MatX, founded by a former senior engineer from Google's TPU team, announced the completion of a $500 million (approximately ¥3.445 billion) Series B funding round.
This financing round boasts an impressive roster of investors. It features strategic participation from leading semiconductor companies like Alchip and Marvell, alongside substantial investments from several top-tier venture capital firms.
Core Weapon: The MatX One Chip
MatX's confidence in this funding round is anchored in its next-generation processor currently in development: the MatX One. This chip is designed to address the dual challenge of achieving both "high throughput" and "low latency" in large model inference.
Innovative Architecture: It employs a "partitionable systolic array" design. This structure ingeniously merges the ultra-high energy efficiency of a large array with the scheduling flexibility of smaller arrays, optimizing hardware utilization.
Memory Breakthrough: The MatX One integrates an ultra-low-latency SRAM architecture with the extended context-handling capacity of High Bandwidth Memory (HBM), overcoming traditional memory bottlenecks in chip design.
Full-Scenario Adaptability: From foundational prefill tasks and high-frequency decoding to complex reinforcement learning training, the MatX One aims to deliver industry-leading performance across all workloads.
Business Outlook: Reducing LLM Operational Costs
A primary goal for model developers in today's compute market is reducing the cost per generated token. Citing MatX's official statement, IT Home reported that the company's technology has the potential to achieve throughput efficiency on par with or exceeding conventional chips, thereby significantly lowering the barrier for deploying and maintaining large-scale models.
Industry Overview: The AI Chip Arms Race Heats Up
The rise of MatX is just one part of the broader global surge in AI chip development. Recent industry activity has been intense:
SambaNova unveiled its fifth-generation RDU chip and entered a deep strategic partnership with Intel.
Positron announced its Asimov chip, claiming its performance per watt can reach five times that of NVIDIA's upcoming Rubin architecture.
Domestic Breakthrough: A research team in China recently developed a flexible AI chip with a production cost under $1. Capable of withstanding 40,000 bending cycles, it points toward new possibilities for wearable AI hardware.
Apple Smart Glasses Could Debut at WWDC27, Highlighting Privacy Protection
Bloomberg’s Mark Gurman reports that Apple’s smart glasses, codenamed N50, are slated for a WWDC27 debut in June 2027, with a retail launch expected in autumn 2027. Originally targeted for late this year and early 2027, the device’s release has been
Inside Details Exposed About Next-Gen Gemini: Strained Computing Power, Internal Teams Disagreed on Development Priorities and Resource Allocation
Reports indicate that the launch of Google’s highly anticipated next-generation Gemini model has been pushed back. Internal disagreements over development priorities and resource allocation, combined with limited computing capacity and complex approv
OpenAI Dismisses Growth Slowdown Concerns, Says Multiple Business Units Accelerating
In response to external scrutiny regarding decelerating sales growth and missed internal benchmarks, AI leader OpenAI issued a confident statement on Tuesday, April 28. The company clarified that its consumer products and enterprise services are adva





Home






