• Arm Chips Are Back, and This Time They Mean Business

    Arm AGI server CPU beauty shot

    Arm has entered the chip market with the AGI processor. Featuring 136 Neoverse cores and PCIe Gen6, it aims to outperform x86 servers by 2× per rack. continue reading

  • What the Heck is a Groq?

    Image of Nvidia Rubin GPU and Groq LP30 LPU

    Groq was one of many companies to realize that GPUs suboptimally execute neural networks. But like other NPU companies, it also failed to anticipate how AI models were changing. Groq’s technical shortcomings, along with the software and other challenges NPU startups face, diminished its products’ appeal to most customers. In a reversal, the strength-weakness duality… continue reading

  • Meta Bares MTIA Roadmap, Accelerates NPU Development

    picture of an MTIA-based system

    Tamping down speculation about the viability of its MTIA AI accelerator, Meta has disclosed its roadmap for the NPU family. The company is speeding up the pace at which it introduces new versions, revealing plans to deploy three designs this year and next after recently deploying its third-generation MTIA300. We believe the first MTIA (now… continue reading

  • Byrne-Wheeler Report Discusses AI Deals, Broadcom and Nvidia Earnings

    BWR 10 thumbnail: Joe, Bob, some text, a background image

    Joe and Bob discuss deals (Nvidia-Meta, AMD-Meta, Amazon-OpenAI, Nvidia-OpenAI), AI-chip startup funding (MatX, Taalas, HyperAccel, SambaNova, Ayar Labs), and earnings announcements (Nvidia, Broadcom). continue reading

  • New Funding and Chip Whisk SambaNova to its Next Stage

    SambaNova SN50 chip

    Hoary startup SambaNova has closed a $350 million Series E funding round, introduced the SN50 AI accelerator, and signed up SoftBank to deploy the new NPU. Following reports that Intel was considering acquiring the company, which Intel CEO Lip-Bu Tan chairs, the beleaguered giant merely joined the funding round and signed a collaboration agreement. SambaNova… continue reading

  • HyperAccel Promises Power Efficiency Gains

    Creative rendering of Bertha NPU

    HyperAccel seeks to displace Nvidia by delivering a lower-cost AI accelerator. Its Bertha 500 promises to double performance while raising power efficiency 12 × compared with a Hopper-generation Nvidia GPU. Based on the company’s latency-processing unit (LPU) architecture, the startup’s NPU achieves these gains through its 90% compute-unit utilization during inference. Bertha’s efficiency is a… continue reading

Popular

Sponsored Content and Plugs

Other Sites You Might Like

Read More

Amazon AMD Arm auto Broadcom business Cerebras Ceva consumer CPU data center DPU DSP edge AI embedded Epyc FPGA Google GPU Imagination Intel interconnect Marvell MCU MediaTek memory Meta MEXT Microsoft MLPerf NPU (AI accelerator) Nvidia OpenAI PC process tech Qualcomm Renesas RISC-V SambaNova Semidynamics SiFive smartphone software Tenstorrent Upscale

error: Selecting disabled if not logged in