ai-chips

AMD has reached a definitive agreement to acquire Toronto-based AI chip startup Taalas, which specializes in hardwiring AI models onto custom silicon for efficient inference. The deal aims to strengthen AMD's AI portfolio with differentiated inference performance and efficiency.

Anthropic is building a team to design its own custom AI chips, aiming to co-design hardware and models for greater efficiency as demand for Claude rises. The move follows OpenAI, Google, and Meta in pursuing in-house silicon.

Lamb Labs (YC S26) announces custom AI inference chips claiming 20,000+ tok/s and 63x higher Intelligence per Watt than traditional GPUs, arguing GPUs were adopted for availability rather than suitability.

An NYT analysis details the unprecedented global build-out of AI data centers and chips, projecting a tenfold increase in AI computing power by 2028, which is expected to drive major breakthroughs in AI capabilities.

Samsung chip engineers are defecting to rival SK Hynix, attracted by larger bonuses fueled by profits from AI memory chips. The talent war highlights the fierce competition in the high-bandwidth memory market.

This article discusses how next-generation AI chip startups are attacking the data movement aspect from different angles in an attempt to challenge Nvidia's dominance, citing a list from July 2026.

Nvidia CEO Jensen Huang and his wife donated $75 million to Vanderbilt University to establish a new college of art, architecture and design in San Francisco, arguing that art determines the purpose of technology.

AMD and Anthropic signed a deal for tens of billions of dollars' worth of AI servers, with Anthropic purchasing up to 2 gigawatts of AMD's Instinct MI450 chips and AMD investing up to $5 billion in Anthropic.

Anthropic and AMD have agreed on a deal worth tens of billions of dollars: Anthropic will buy up to 2 gigawatts of AMD's AI chips, and AMD will invest up to $5 billion in Anthropic.

Alibaba's T-Head open-sourced SAIL, the software stack for its Zhenwu AI chips, at WAIC in Shanghai. The move aims to lower the barrier for developers to migrate from Nvidia's CUDA ecosystem.

SemiAnalysis argues that Kimi K3's linear attention (KDA) is not detrimental to NVIDIA, HBM, DRAM, and networking, contrary to uninformed panic, and explains why reduced KV-cache requirements are actually beneficial.

XPeng announced its new budget EV, the L03, at a Munich showcase event, targeting global markets with a starting price of €35,600 and features like fast charging, a 320-mile range, and AI-powered systems.

This article explains how systolic arrays handle over 95% of AI chip compute, detailing their design, modes of operation, and why they are efficient for matrix multiplication.

Apple's failed self-driving car program led to the development of the Neural Engine and powerful on-device AI chips, which now underpin Apple's AI hardware strategy, including upcoming M7 chips with significant Neural Engine upgrades.

SK Hynix raised $26.5B in the largest foreign IPO in US history, driven by demand for its HBM memory chips used in AI GPUs. The US Commerce Secretary urged the company to build new fabrication plants in the US to reduce reliance on South Korean production.

Meta's custom AI chips (MTIA) will begin production in September, aiming to reduce GPU costs. The chips are designed with Broadcom and manufactured by TSMC, part of Meta's strategy to secure compute capacity while still purchasing from Nvidia and AMD.

A newsletter covering a US nuclear milestone with four microreactors achieving criticality, and China planning to let its top AI firms purchase Nvidia H200 chips, alongside other tech stories.

SambaNova raises $1 billion at an $11 billion valuation to challenge Nvidia in AI inference chips, with JPMorgan deploying its on-premise systems and an IPO likely in 2027.

Facing US export controls, Chinese AI startup DeepSeek is developing its own inference chips to reduce reliance on both Nvidia and Huawei, joining a trend of AI companies designing custom silicon.

A thread analyzing how six AI chip competitors (Tenstorrent, Cerebras, Trainium, TPU, SambaNova, Furiosa) all independently abandoned traditional GPU features like hardware caches and threads, using software-managed SRAM and different programming models, contrasting with NVIDIA's CUDA approach.