Hut 8 priced $4.25 billion of 6.129% senior secured notes due 2042 to finance its Beacon Point AI data center development in Nueces County, Texas. The planned campus is designed for 352MW of critical IT capacity across six data halls, and Hut 8 said it will be leased to a tenant rated AA- or higher. The offering is set to close on June 9, 2026, with interest beginning Nov. 30, 2026 and amortization starting May 30, 2030.

Leopold Aschenbrenner expanded Situational Awareness LP from $255M to $13.7B over six quarters, with Q1 2026 showing the largest step-up. The filing shows a new $8.5B put-options overlay led by a $2.04B SMH put and additional puts on names including NVDA and AMD, while several long positions in AI-power and crypto-mining themes were increased.

Marvell Technology will be added to the S&P 500 on June 22 after meeting the index's GAAP profitability requirement and will replace Pool Corp. The change compels funds that track the benchmark to purchase Marvell shares, after the stock jumped nearly 6% in after-hours trading following the June 5 announcement. The reshuffle arrives as chip stocks saw a sharp selloff, with the PHLX semiconductor index down 10.3% in the same week.

Goldman Sachs projects SpaceX's AI-related revenue could climb to $322Bn by 2030 from an estimated $3.2Bn in 2025, with total company revenue modeled at $474Bn. The report ties the thesis to Starlink-enabled "orbital compute" and comes as SpaceX targets a $75Bn IPO raise at a $1.75 trillion valuation, with pricing expected June 11 and trading slated for June 12.

A factory in Guangdong can assemble one humanoid robot every 30 minutes, as China accounted for nearly 90% of global humanoid robot sales in 2025. Global deliveries topped 13,000 units in 2025, while leading Chinese vendors Unitree and Agibot shipped 5,500 and 5,168 units, respectively. Cost comparisons suggest the West may pay far more for comparable builds, largely due to supply-chain dependence on Chinese components.

Apple is preparing to power its redesigned Siri with Nvidia Blackwell B200 processors running inside Google's data centers, shifting away from its earlier plan to keep cloud AI requests solely on its own Private Cloud Compute servers. The company is expected to outline the changes at WWDC on June 8, with a public release aimed for September alongside iOS 27. Apple also plans to use Nvidia's confidential compute capability to encrypt data while it is processed in the cloud.

On June 2, 2026, President Trump signed an executive order asking AI companies to provide their most advanced frontier models to the U.S. government up to 30 days before public launch. The White House describes the approach as voluntary rather than regulatory, while also calling for "trusted partners" and faster cyber-defense work on critical national-security networks within 30 days.

The global semiconductor space has found itself at a crucial macro crossroads. For years, the core debate surrounding artificial intelligence scaling sat squarely on a single question: "Who has the graphics processing units (GPUs)?" However, a deeper shift is taking place as the market moves from speculative infrastructure build-outs to local processing, full-stack software integration, and self-contained computing units.