Elon Musk’s SpaceX is pressing ahead with the idea of putting AI data centers into orbit, arguing it could become more practical as launch costs fall and terrestrial buildouts face growing local pushback. The company’s IPO raised $85.7 billion and valued the newly public SpaceX in the trillions, giving it new capital to link rockets, Starlink satellites and xAI’s computing needs. Still, investors and economists say the economics remain uncertain even as other players, including Jeff Bezos and Google, pursue similar concepts.

Marvell $MRVL has rallied from a sub-$100 stock to above $300 after Jensen Huang called it “the next trillion-dollar company,” prompting investors to search for his next signal. The keynote’s most actionable clue was a single slide laying out an “AI factory” ecosystem map. The guest post is from Ziven.io, a public markets intelligence platform, the article said.

Big Tech stock buybacks are falling as companies prioritize spending on artificial intelligence, shifting from capital-light approaches to more capital-intensive strategies. Among Alphabet, Microsoft, Meta and Amazon, only Microsoft repurchased shares in the first quarter. Microsoft’s $3.4 billion buyback was the smallest total within the group in nearly a decade, according to data compiled by Bloomberg.

Abu Dhabi-backed AI investor MGX is weighing a multibillion-dollar acquisition of DayOne, which operates a data center in Singapore. The deal would be MGX’s first acquisition in Asia. It would underscore intensifying competition to secure the data center infrastructure needed to support AI growth.

Japan on June 19 outlined a plan to direct 10.5 trillion yen ($65.1 billion) in combined public and private funding into physical artificial intelligence across 17 strategic sectors by fiscal 2040, according to Nikkei Asia. The initiative is intended to address labor shortages tied to Japan’s shrinking workforce, including in manufacturing, logistics, healthcare, and eldercare. It is also the fifth major policy move under Prime Minister Sanae Takaichi’s government within four days.

Microsoft is carving out an unusual role in the U.S.-China AI rivalry by selling OpenAI’s GPT models to major Chinese internet companies through its Azure cloud service. With OpenAI and Anthropic declining to enter China directly, firms such as ByteDance, Ant Group and Tencent can still access their technology via Microsoft. At the same time, some U.S. developers are moving to lower-cost Chinese models such as DeepSeek. Beijing is also rolling out new policies aimed at speeding AI adoption across the broader economy.

A proposal dubbed Root Reborn would reshape how Bittensor validators handle yield by reinvesting subnet proceeds instead of automatically selling them. Under the plan, validators would select baskets of subnets to support, resembling portfolio managers, while stakers could still convert returns back to TAO when they choose. The change would shift subnet economics by replacing persistent sell pressure with reinvestment flows.

Solana’s engineering team released new consensus research that challenges the traditional “leader handoff” design and explores an architecture that does not require each block to reference its immediate parent. In a 10-node test cluster spanning Asia, Europe, and North America, the team demonstrated 10ms slot times with block leaders rotating every 10ms despite cross-region delays exceeding 100ms. The work is positioned as research only and does not change the project’s direction for Constellation, but it shows a potential path to moving beyond the network-delay Δ constraint.