Blue Origin's First Outside Round Oversubscribed as AMD Pays $8.2B for World Labs

Blue Origin's first ever outside capital raise is not closed, and investors are piling in. CEO Dave Limp said so on Tuesday at the Trump administration's "Hello, America" tech summit in Washington, D.C. He called the round "oversubscribed."

That is a notable shift for a 26-year-old company that Amazon founder Jeff Bezos has funded alone since he started it in 2000. Limp did not confirm the final size of the round or how much Bezos has put in. "I'm so amazed the investor community embraces space this much," he said, according to CNBC, which reported his remarks on 30 September. Last week the Wall Street Journal reported Blue Origin had already raised $10 billion at a $140 billion valuation, with Bezos contributing $2 billion of that and $30 billion invested since founding. Limp declined to confirm those numbers. The gap matters. An oversubscribed round can still close higher or lower than the figure a paper reported before the book was shut.

Space capital is not the only money moving

The same week brought a much larger and more concrete transaction in AI. AMD and World Labs said AMD will acquire the world models company by the end of the year, pending regulatory approval, in a deal valued at $8.2 billion. Ars Technica reported the agreement on 29 September.

World Labs was founded in 2024 by computer vision scientist Fei-Fei Li with Justin Johnson, Christoph Lassner and Ben Mildenhall, on $230 million in funding, some of it from AMD. Its models aim to simulate the physical world well enough to be useful for prediction. The company's first public tool, Marble, generates small 3D spaces using Gaussian splats that can be exported as assets for film and game production. Later models and tools, including Atlas, shifted emphasis. The commercial pull behind the sector is synthetic data for robot training, a field where Nvidia, AMD's direct competitor, holds a strong lead. AMD has shipped its own AI models, including the open source Micro-World, but Ars Technica notes it has not produced a stack that competes with Nvidia's Cosmos line. Under the deal, Fei-Fei Li becomes executive vice president and chief scientist at AMD, while Johnson and Mildenhall continue leading the World Labs team as it joins AMD. AMD's announcement frames the purchase around reasoning, robotics, simulation and physical AI.

Two deals, two very different funding moments. One is a private round that has not priced. The other is a signed acquisition with a number attached.

Bezos's rocket company and its repair job

Blue Origin's raise lands against an operational backdrop that is anything but settled. In late May, New Glenn suffered a catastrophic explosion on the company's Cape Canaveral launch pad, causing extensive damage. Limp said on Tuesday that the plan is still for New Glenn to return to flight before the end of this year, with 1,000 people working the pad, 500 on day shift and 500 on night shift. The rocket's record is mixed. New Glenn reached orbit on its first flight early last year and landed its booster on the second mission last fall. A third mission in April saw the booster land, but the payload, a satellite for AST SpaceMobile, deployed in the wrong orbit and was lost.

Investors are also being asked to price two satellite constellations that do not exist yet. TeraWave, unveiled in January, is a planned communications service for enterprise, data center and government customers, not consumers, with more than 5,400 satellites in low Earth orbit. Project Sunrise, disclosed in March, would deploy about 56,000 satellites capable of handling AI computing from space.

Limp gave some of his most detailed comments yet on the orbital data center strategy. He said many heavy data centers would go into sun-synchronous orbit where solar power is constant, launched on New Glenn. Those centers would have limited communications capability and connect into TeraWave to deliver results to Earth, which Limp said would enable access to "effectively infinite compute." The company is targeting a launch cadence ramping from hundreds to thousands of missions a year. Asked why Blue Origin is engaging outside investors now, Limp cited a "combination of things": New Glenn's flights, the two constellation announcements, and the need to attract and retain staff. "Attracting great talent … is difficult in a competitive world, and obviously having an external valuation on the company that shows employees what the value is is very helpful," he said. He added: "I'm of the firm belief this will be the best investment Jeff has ever made, and by the way, he has some good ones on his track record."

SpaceX's IPO in June gave the sector a public benchmark and pulled more attention to commercial space. Blue Origin's round is the private-market echo of that.

Where the newest money is going

AI remains the dominant destination. AMD's $8.2 billion cheque for World Labs, a company founded barely two years ago, is the clearest signal this week of what chipmakers will pay to close a gap with Nvidia. Ars Technica notes the acquisition came sooner than expected for such a young company, though the prior AMD investment makes it less of a surprise.

Nuclear and fusion startups are pulling capital for a different reason: power for data centers. NPR reported on 30 September that Valar Atomics is planning "Project Beehive" on more than 9,000 acres of Bureau of Land Management land near Price, Utah, with data centers and about 456 small nuclear reactors, plus facilities to produce nuclear fuel and store waste. The documents were reviewed exclusively by NPR. The aggregate output would be around 9.6 gigawatts for the data centers, and more thermal heat for industrial uses. For scale, NPR notes the entire state of Utah produces only around 4 gigawatts of electricity on average, with summer peak capacity near 10 gigawatts. The timeline calls for nonnuclear construction to begin as soon as the end of this year, with the first reactors online sometime in 2028. Valar has also sought around 650 acres of state land; the Salt Lake Tribune first reported that state land use last week, but not the larger federal proposal.

The BLM's Utah office confirmed to NPR it had received an application and is reviewing it for completeness. Valar said no construction begins before permits and approvals. Local groups are not reassured. "It's concerning to see that the Trump administration, with the support of the state of Utah, is pushing this forward on significant acreage of public lands," said Neal Clark, wildlands director at the Southern Utah Wilderness Alliance. Lexi Tuddenham, executive director of the Healthy Environment Alliance of Utah, said the biggest thing her group can do right now is let people know about it.

Fusion, the other side of the nuclear bet, is slower. Japan's Helical Fusion expects to begin preliminary power-on tests at its demonstration reactor in 2027, the company said on Tuesday, according to The Asahi Shimbun. The Helix HARUKA pilot at the National Institute for Fusion Science in Toki, Gifu Prefecture, is Japan's first integrated pilot reactor using the helical approach. Construction of the first spiral coil started early this year and is now largely complete. If first-stage tests succeed, Helical Fusion plans a second coil and other major components, including a liquid-metal wall blanket, before power-on tests of the completed facility around 2030. A larger reactor, HELIX KANATA, is planned with a net output of 50 MW, with full-scale investment in the 2030s. CEO Takaya Taguchi told reporters the company aims to build a commercial plant as early as the 2040s, but said cost, safety and construction speed remain hurdles alongside the technical ones. Japan has designated fusion one of 17 strategic sectors and plans 3.1 trillion yen ($19.7 billion) in public and private funding by the 2040 financial year.

Seventy years of fusion research have yet to produce a commercially viable reactor, as the Asahi report notes. That does not stop money from arriving.

Take the four data points together. Blue Origin: round open, size unconfirmed, valuation reported by the Wall Street Journal at $140 billion. AMD and World Labs: $8.2 billion, pending approval, signed. Valar Atomics: 456 reactors proposed on federal land, application under review. Helical Fusion: tests in 2027, commercial plant maybe in the 2040s. Only one of those is a closed transaction, and it is the one with the smallest chance of being repriced.