xAI signed a compute agreement with Anthropic for access to its Colossus 1 supercomputer in Memphis, which runs more than 220,000 NVIDIA GPUs across H100, H200, and GB200 deployments and adds over 300 megawatts of capacity to Anthropic's stack within the month. The capacity will be applied directly to Claude Pro and Claude Max subscribers. The agreement separately discloses that Anthropic and SpaceX have expressed interest in jointly developing multiple gigawatts of orbital AI compute capacity, citing terrestrial power, land, and cooling constraints on next-generation training and inference.

For SpaceX investors, the orbital work stream is the more important half of the announcement. Anthropic is the first frontier AI lab to name SpaceX as a development partner for orbital compute in a written commercial agreement rather than a research note. This supports our thesis that SpaceX has the optionality as both a captive and merchant provider of orbital data center. The Mach33 view is that SpaceX captures the orbital compute market before that market formally exists, because every orbital compute customer has to buy launch, bus, and on-orbit operations from someone, and SpaceX is the only vendor with the vertical integration at the scale this thesis requires. 

Local real estate insider Jim Keaty of Keaty Real Estate and multiple Vermilion Parish sources report credible but unconfirmed indications that SpaceX may be acquiring approximately 136,000 acres of coastal marshland near Pecan Island and Freshwater City in Louisiana. Supporting signals include cancelled hunting leases for the entire 2026 season, unsolicited offers from out-of-state speculators at three to ten times appraised value, and ExxonMobil's stalled wetlands permits for a 125,000-acre carbon capture project in the same footprint. The proposed campus would reportedly focus on manufacturing, testing, and barge logistics via the Gulf Intracoastal Waterway rather than new launch pads. Status is market rumour with no SpaceX confirmation or regulatory filings to date.

If confirmed, this would be among the largest land deals in SpaceX's history, dwarfing Starbase Texas in raw acreage, and would slot into the company's vertical integration playbook in a way that directly supports Starship production cadence between Boca Chica and Florida. The barge logistics angle fits Starship's road-impossible dimensions and the halfway-point geography of Vermilion Parish along the GIWW, while the ExxonMobil permitting stall suggests the seller side has motive rather than just speculative inquiry. Sophisticated investors should treat this as a pre-IPO infrastructure scaling rumour worth tracking through Vermilion Parish records, FAA environmental filings, and the forthcoming public S-1, with credible read-through to long-term Starship cadence assumptions and the IPO narrative around manufacturing capacity.

STMicroelectronics held an investor conference call on May 4 in which Remi El-Ouazzane, President of the Microcontrollers, Digital ICs and RF Products Group, guided to over $3 billion in cumulative space semiconductor revenue from 2026 to 2028 and a serviceable LEO broadband electronics market growing from $650 million in 2025 to roughly $2 billion in 2028 and close to $3 billion by 2030. He cited Gartner's expectation of close to $15 billion in global LEO services spend in 2026. LEO revenue rose from $175 million in 2021 to roughly $600 million in 2025 and approaches $1 billion in 2026. STMicroelectronics holds a near 90% share in LEO RF semiconductors, has shipped over 7.5 billion integrated circuits to the Starlink programme, and reports that bill-of-materials content per satellite has grown 8x between Starlink V1.5 and V3 to several tens of thousands of dollars. Per ST investor call and Reuters wire reporting.

The SAM trajectory is the most aggressive electronics-market sizing yet from a tier-one Starlink supplier and frames a market roughly 5x the size of ST's current LEO revenue base by 2030. For SpaceX the implication is twofold: ST chip content of several tens of dollars per terminal supports a high-volume consumer manufacturing model and the supplier validation hardens the 2026 to 2028 ARPU and churn modelling that pre-IPO secondary buyers are using. El-Ouazzane offered a wild guess of three years until a relevant volume of orbital data centres reach orbit, which on Mach33's read aligns with the ODC milestone in Musk's compensation plan and with Starcloud's funding trajectory. Material implications for SpaceX revenue ramp scenarios, IPO timing pressure, and competitive positioning against Amazon Leo in the same growth window.

Starcloud is in talks to raise at least $200 million at a $2.2 billion post-money valuation, exactly double the $1.1 billion mark set by its Series A roughly one month earlier. The Information broke the story on May 1, with SpaceNews independently confirming via a separate single source. Starcloud is developing a proposed 88,000-satellite constellation.

The near-vertical valuation step within roughly thirty days is the strongest signal yet that institutional capital is repricing the orbital compute thesis rather than treating it as a speculative niche. For SpaceX the read-through is twofold: Starcloud is a probable Starship customer at scale, and the round validates the ODC milestone now embedded in Musk's compensation plan as a credible value driver rather than long-horizon optionality. Sophisticated investors should watch for secondary market repricing of related ODC plays, including Aetherflux, and the broader category of orbital infrastructure entrants raising at billion-dollar marks despite no operational hardware in orbit.

Blue Origin posted job openings in late April 2026 signalling plans to ramp New Glenn upper stage production from 12 to 60 units per year by Q3 2028, with longer-term targets reaching 100 annually by 2029. The hiring push comes as the rocket remains FAA-grounded following the April 19 NG-3 mission, in which a reused first stage successfully landed but a second-stage anomaly placed AST SpaceMobile's BlueBird 7 into an off-nominal orbit, leading to deorbit and an insurance write-off. Production targets are signalled rather than confirmed and remain contingent on mishap closure.

The cadence ambition would put Blue Origin within structural striking distance of competing for Amazon Leo and national security heavy-lift contracts at scale, but the gap between ambition and delivery is the dominant variable. Investors should weigh the first booster reuse demonstration as a technical milestone against the upper-stage failure mode that triggered the grounding, and against a production target curve that requires 5x growth by Q3 2028. For AST SpaceMobile, multi-launcher diversification reduces single-point exposure but compresses 2026 service ramp timelines, while SpaceX stands to absorb residual heavy-lift demand during any extended grounding.

SpaceX's confidential draft registration statement, reviewed by Reuters, contains the first public figure on Starship programme spend: more than $15 billion to date, against the roughly $400 million SpaceX spent developing Falcon 9 to operational status (rather than develop it from scratch; there were many predecessors to Falcon 9). The filing also indicates SpaceX targets V3 Starlink deployment in the second half of 2026, with Starship's payload bay configured to carry up to 60 V3 satellites per flight. The filing also shows SpaceX directed all $3 billion of 2025 space-segment R&D into Starship, up from $1.8 billion the prior year. All figures sourced from Reuters' review of the confidential filing.

Pre-IPO secondary tenders are likely to reference the $15 billion figure as a floor for what new entrants would need to match, which hardens SpaceX's moat against Blue Origin, Rocket Lab Neutron, and Stoke Space. This is also a reminder that the $1.75 trillion IPO target prices successful Starship reusability rather than current test cadence.

NASA plans to raise the maximum value of its Commercial Lunar Payload Services programme from $2.6 billion to $4.2 billion, per a procurement filing posted to SAM.gov on April 27. The $1.6 billion uplift, a 62% increase, supports the higher robotic-lander cadence required by NASA's Moon Base programme (Project Ignition), announced in March. Intuitive Machines was selected on March 24 for IM-5, a $180.4 million task order to deliver seven payloads to Mons Malapert near the lunar South Pole, targeting a 2030 launch.

The contract uplift is the strongest signal to date that CLPS is moving from experimental to recurring revenue, providing forward visibility for Firefly Aerospace, Intuitive Machines, and the broader commercial lunar supply chain. The headroom is contractual, not yet flow: task orders awarded to date total under $2 billion across the programme's life, and at NASA's Lunar Surface Innovation Consortium meeting on April 29, industry representatives said they could scale production but stopped short of committing to the rates implied by the agency's projections. Joel Kearns, NASA's deputy associate administrator for exploration, said on the same panel he was not familiar with the filing.

The FCC unanimously adopted Report and Order FCC 26-26 on April 30, replacing the legacy EPFD framework with performance-based GSO protection criteria. The Commission's own analysis estimates the change could unlock up to 7× additional capacity for non-geostationary operators and generate over $2 billion in U.S. economic benefit. Confirmed via FCC press release and adopted order.

The 7× headline overstates the discrete event. SpaceX has been operating under a comparable interim regime since January 9 via waiver DA 26-36 (Kuiper received the parallel waiver February 10), so most of the capacity lift was already in production before the vote. What changed April 30 is durability: removal of expiration risk, replacement of fixed EPFD caps with performance-based backstops (3% throughput degradation, 0.1% link unavailability), and reduction of the GSO arc avoidance angle from 4° to 3°.

The economic value lands in V3, not V2 Mini. Per Mach33 modelling, post-vote rules combined with the V3 deployment ramp push the supply-side US TAM from roughly 8.6 million addressable households today to approximately 117 million at full V3 deployment, at which point demand-side economics, not satellite physics, becomes the binding constraint. Suburban regions absorb the bulk of the unlock. Read our weekly analysis that examines this in full.