Tech & Science

Anthropic taps SpaceX data center to ease Claude capacity crunch

Anthropic CEO Dario Amodei revealed Wednesday that the company achieved 80-fold year-over-year revenue growth on an annualized basis in the first quarter o...

Anthropic CEO Dario Amodei revealed Wednesday that the company achieved 80-fold year-over-year revenue growth on an annualized basis in the first quarter of 2026, a figure that far exceeded the company’s internal projections of 10x growth. Speaking at Code with Claude, Anthropic’s developer conference in San Francisco, Amodei attributed much of the explosive demand to the runaway adoption of Claude Code among software developers.

Anthropic taps SpaceX data center to ease Claude capacity crunch

SpaceX Compute Deal

The surge in usage has strained Anthropic’s computing infrastructure, prompting the company to announce an unusual partnership with SpaceX. Under the deal, Anthropic will gain access to the full capacity of SpaceX’s Colossus 1 data center in Memphis, Tennessee, adding more than 300 megawatts of compute power and over 220,000 Nvidia GPUs within the month.

Ami Vora, Anthropic’s chief product officer, announced that the new capacity would allow the company to immediately double Claude Code’s five-hour rate limits for Pro, Max, Team, and Enterprise plans, while also removing a peak-hours reduction that had frustrated paying users. The deal follows similar compute arrangements Anthropic has struck with Amazon and Google/Broadcom.

The partnership drew attention in part because SpaceX is controlled by Elon Musk, whose own AI venture, xAI, is a direct competitor to Anthropic. Wired described the arrangement as “an unexpected turn” in the AI industry’s intensifying race for compute resources.

‘Dreaming’ for AI Agents

Anthropic also introduced a memory-refinement technique it calls “dreaming” for Claude Managed Agents. The feature runs evaluations between work sessions, reviewing past behavior, pruning stale notes, merging duplicates, and resolving contradictions in persistent memory files — a process the company likened to how human brains consolidate information during sleep.

The capability, available as a research preview requiring developer access, is designed to address “memory rot” — a common failure mode in long-running AI agent workflows where accumulated instructions become outdated or contradictory. Developers can invoke the feature manually or allow it to operate as a background process between sessions.

Anthropic’s annualized revenue reportedly grew from $1 billion in December 2024 to $30 billion by the end of March 2026, according to Forbes. The company closed a $30 billion Series G round in February at a $380 billion valuation.

Leave a Reply

Your email address will not be published. Required fields are marked *