NextSilicon unveils processor chip to challenge Intel and AMD

Israeli startup NextSilicon is developing a RISC-V central processor to complement its Maverick-2 chip for precision scientific computing, positioning it against Intel and AMD and in competition with Nvidia’s systems. Sandia National Laboratories has been evaluating the technology as the company claims faster, lower power performance without code changes on some workloads.

NextSilicon, an Israeli startup whose computing chips are being evaluated by U.S. national labs, said it is developing a central processing unit that it hopes will rival Intel and AMD while helping it compete with Nvidia’s systems. The effort complements the company’s flagship Maverick-2 chip, which targets high-precision scientific computing tasks such as modeling nuclear weapons. The company is seeking to fill a gap in a domain once dominated by Nvidia, as Nvidia has focused more on lower-precision workloads such as Artificial Intelligence.

For the first time, NextSilicon disclosed that its forthcoming CPU is based on RISC-V, an open computing standard that competes with Arm and is increasingly used by major chipmakers, including Nvidia and Broadcom. Nvidia often pairs its accelerators with in-house or third-party CPUs, even partnering with companies such as Intel to tighten the integration between processors and accelerators. NextSilicon’s move signals intent to offer a more complete platform for scientific and high-performance workloads.

While the new CPU remains a test chip, Maverick-2 is already in production. NextSilicon claims Maverick-2 can perform some of the same work as Nvidia chips, but faster and using less power, and without requiring software code to be rewritten. Sandia National Laboratories has been evaluating prototype systems built with NextSilicon’s chips for three years, indicating sustained interest from U.S. research institutions in alternatives for precision computing.

James H. Laros III, senior scientist and Vanguard program lead at Sandia National Laboratories, said in a statement that the company’s performance results are impressive, showing promise for advancing computational capabilities without the overhead of extensive code modifications. The combination of a RISC-V CPU effort and a production accelerator underscores NextSilicon’s strategy to challenge established vendors in both the CPU and accelerator markets as scientific computing needs evolve alongside the rise of Artificial Intelligence.

58

Impact Score

Intel unveils massive artificial intelligence processor test vehicle showcasing advanced packaging

Intel Foundry has revealed an experimental artificial intelligence chip test vehicle that uses an 8 reticle-sized package with multiple logic and memory tiles to demonstrate its latest manufacturing and packaging capabilities. The design highlights how Intel intends to build next-generation multi-chiplet artificial intelligence and high performance computing processors with advanced interconnects and power delivery.

Reward models inherit value biases from large language model foundations

New research shows that reward models used to align large language models inherit systematic value biases from their pre-trained foundations, with Llama and Gemma models diverging along agency and communion dimensions. The work raises fresh safety questions about treating base model choice as a purely technical performance decision in Artificial Intelligence alignment pipelines.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.