AI News: May 30, 2026 - Anthropic Is Now the World's Most Valuable AI Company, SpaceX Lists in June

1. Anthropic Closes $65B Series H at $965B - Now the World's Most Valuable AI Company

On May 28, Anthropic announced it had closed a $65 billion Series H funding round at a $965 billion post-money valuation - overtaking OpenAI ($852B) as the most valuable AI company in the world for the first time. The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, each contributing more than $2 billion. Co-leads include Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN. The round includes $15 billion of previously committed investments from hyperscalers including $5 billion from Amazon. Anthropic also disclosed that its run-rate revenue crossed $47 billion earlier in May - up from $380B valuation and approximately $10B ARR just three months ago in February.

The funding announcement came alongside the launch of Claude Opus 4.8 - 69.2% on SWE-Bench Pro (vs 58.6% for GPT-5.5), 1890 Elo on GDPval-AA (+121 over GPT-5.5), pricing unchanged at $5/$25 per million tokens. New capabilities include Dynamic Workflows in Claude Code (hundreds of parallel subagents for codebase-scale tasks), a 2.5x Fast Mode at $10/$50, and alignment improvements making the model roughly 4x less likely to let code flaws pass unremarked. Anthropic also confirmed Claude is now the first frontier model available across all three major cloud platforms: AWS, Google Cloud, and Microsoft Azure. TechCrunch described this round as likely Anthropic's final private fundraise before its targeted October 2026 IPO.

Full review: Claude Opus 4.8 benchmarks, pricing, Dynamic Workflows, and the Series H breakdown ->

2. SpaceX IPO: Roadshow Starts June 8, Listing Targeting June 18-30 Under SPCX

SpaceX filed its public S-1 registration statement with the SEC on May 20, 2026, formally kicking off what analysts expect to be the largest IPO in US history. The company is targeting a $1.75 trillion to $2 trillion valuation with a raise of approximately $75 billion - 30% of which has been earmarked for retail investors, three times the typical mega-IPO retail allocation. The investor roadshow is confirmed to start June 8, with listing targeting June 18-30 under the ticker SPCX on Nasdaq. Goldman Sachs, Morgan Stanley, and Bank of America are lead underwriters.

The S-1 financials reveal a structurally complex company. Q1 2026 revenue on a consolidated basis was $4.694 billion, with a $1.943 billion operating loss and $1.127 billion adjusted EBITDA. Full year 2025 revenue was $18.674 billion. The Space and Connectivity segments (rockets and Starlink) are profitable on both an operating and EBITDA basis. xAI - merged with SpaceX in February 2026 at a combined $1.25 trillion valuation - is the primary source of losses. Starlink has 10.3 million subscribers across 164 countries as of March 31, 2026, operating approximately 9,600 satellites in low-Earth orbit. SpaceX holds 18,712 bitcoin worth approximately $1.3-1.5 billion. Elon Musk owns 42% of equity and controls 85% of voting power through a dual-class share structure - meaning public investors will have minimal governance influence over the company they are buying into.

The Anthropic connection is material: SpaceX's Colossus 1 data center (220,000 Nvidia GPUs, 300MW) has a contract with Anthropic worth $1.25 billion per month through May 2029 - approximately $40 billion over the life of the deal - though either party can terminate with 90 days' notice. SpaceX also entered a $10 billion collaboration with Cursor that includes an option to acquire Cursor for up to $60 billion. For AI infrastructure investors, the SpaceX IPO is as much an AI compute infrastructure story as it is a rockets and satellites story.

3. Span and Nvidia Are Putting Mini Data Centers in Suburban Backyards

California startup Span has partnered with Nvidia and homebuilder PulteGroup to mount XFRA nodes - compact units with 16 Nvidia RTX Pro 6000 Blackwell GPUs, 4 AMD EPYC CPUs, and 3TB RAM - on the exterior of newly built homes, using unused household electrical capacity (the average US home uses only 40% of available grid capacity) to create a distributed AI compute network. The units are roughly the size of an HVAC condenser, liquid-cooled and fanless to eliminate noise. Homeowners receive a flat fee plus compensation based on energy and network use.

Span claims the approach is 6x faster to deploy and 5x cheaper than a traditional 100-megawatt data center. One home is currently live. A 100-node Q3 2026 pilot is confirmed for a southwestern US state - Nevada or Arizona likely. The 2027 target is 80,000 nodes delivering over 1 gigawatt of combined compute. Nvidia's Marc Spieler framed the strategic logic: "We're trying to get access to power, and there's a lot of power right now on the grid. But unfortunately, to come up with large loads for big data centers - it's a challenge." PulteGroup operates 1,043 active communities across 45+ US markets, giving the partnership direct access to new construction at national scale.

Full breakdown: Span XFRA - what each node contains, the economics, the pilot timeline, and the honest challenges ->