compute

A commenter questions why China, despite securing massive compute resources, appears to still rely on distilling US AI models rather than training foundational models from scratch, wondering what the actual bottleneck is.

An observation that future technologies often start as seemingly trivial toys before evolving into foundational infrastructure for civilization.

Observations of a shift where large enterprises are increasingly seeking to secure compute and post-train their own models in-house, often on open-source GLM-5.2, highlighting the growing acceptance of open-source AI.

SpaceX acquires AI coding platform Cursor for $60 billion in an all-stock deal, aiming to leverage xAI's compute infrastructure and compete with Anthropic and OpenAI.

A Hacker News discussion explores whether Europe can train a frontier AI model using its own compute resources, highlighting the region's AI sovereignty challenges.

Analyzes the argument that AI compute cannot become a commodity due to non-fungibility, using commodity market analogies to suggest that a standardized reference plus basis trading could enable commoditization.

The article discusses a proposed coalition to build AI infrastructure in the free world, focusing on compute resources and policy implications.

Anthropic is moving toward a public IPO filing after raising $65 billion at a $965 billion valuation, with co-founder Daniela Amodei citing the need for capital to train frontier models and serve inference at scale. The company reported $47 billion in annualized revenue as of May 2026, and has partnered with xAI for $1.25 billion per month in compute capacity rather than building its own data centers.

Alphabet announces an $80 billion equity capital raise to expand its AI infrastructure and computing capabilities, signaling a massive commitment to advancing AI technology.

A comment highlighting that replacing human workers with AI agents will require far more compute than commonly assumed, as reliable agents need extensive tool use, self-checking, and coordination.

Surprising new results show that for large LMs with enough compute, the best data filter might be no filter, as they tolerate low-quality data well.

Anthropic agrees to pay SpaceX nearly $45 billion over three years for compute resources, expanding an earlier partnership that included capacity at SpaceX's Colossus 1 data center.

Anthropic will pay xAI $1.25 billion per month for compute capacity from the Colossus 1 data center, in a deal that could bring xAI over $40 billion in revenue.

Sam Altman announces OpenAI's Guaranteed Capacity, offering discounted tokens for 1-3 year commitments to provide customers with capacity certainty.

OpenAI announced Guaranteed Capacity, a new offering that provides customers with long-term guaranteed access to compute via 1-3 year commitments and discounts, enabling reliable scaling for critical workloads.

Leopold Aschenbrenner backs crypto miners pivoting to AI infrastructure, leveraging their existing expertise in power, cooling, and large-scale facilities for AI data centers.

crunr is a tool that allows users to launch and run any compute job on AWS with a single command, simplifying cloud compute management.

Miles Brundage argues that automated AI R&D and recursive self-improvement are fundamentally dependent on compute being above a critical level and continuously increasing.

The author proposes using AI to scan signals from power grids, datacenters, and other sources to extract changes in power dynamics and generate better questions about the AI economy, rather than just answering existing questions.

Anthropic CFO Krishna Rao discusses the company's revenue growth from $250 million to $30 billion, fundraising of ~$75 billion, compute procurement, and use of Claude by the finance team in a recent podcast appearance.