Share with your CTO
SpaceX’s first post-IPO earnings report confirmed $15.83 billion in AI compute capital expenditure in a single quarter, and the company named NVIDIA the exclusive chip supplier for its Starmind orbital compute program. NVIDIA’s Vera Rubin architecture got the socket; AMD, despite reporting 50% year-over-year revenue growth and a doubling of data center revenue to $6.72 billion, lost the design win and traded down nearly 7%. The market’s verdict was swift: a lost exclusive supply relationship erases a record quarter, full stop.
What this means for your business
The enterprises most exposed to this story aren’t trading ETFs. They’re the ones mid-cycle on an AI infrastructure roadmap where AMD was penciled in as a competitive check on NVIDIA pricing. SpaceX’s exclusive commitment to Vera Rubin doesn’t just hand NVIDIA a trophy; it tightens NVIDIA’s negotiating leverage with every hyperscaler and large enterprise that hoped AMD’s data center momentum would create a credible alternative. If you were counting on AMD’s trajectory to keep your GPU procurement honest, that thesis just got repriced.
The $15.83 billion single-quarter figure matters beyond the headline. NVIDIA guided to $91 billion in revenue for its current quarter, and CEO Jensen Huang’s framing of “the largest infrastructure expansion in human history” now has a concrete demand signal behind it, not just a narrative. One customer spending that much in ninety days on NVIDIA-anchored compute confirms the demand curve isn’t slowing, which means the capacity constraints and extended lead times that made GPU procurement painful in 2024 and 2025 aren’t structural anomalies. They’re the baseline condition for anyone building serious AI infrastructure.
AMD’s position deserves a sharper read than the stock move suggests. Record revenue, doubled data center sales, and a $13 billion forward guidance number all got discounted because the market cares about design wins at the frontier, not trailing revenue. The practical implication for enterprise architects is that AMD remains a real option for inference workloads and cost-sensitive deployments, but any internal business case that relied on AMD closing the gap with NVIDIA at the bleeding edge of training and orbital compute should be revisited. The gap didn’t close today; it got a high-profile public stamp confirming it’s still there.
Concept deep-dive: Leveraged daily-reset ETFs
A daily-reset leveraged ETF delivers 2x (or inverse 2x) of its underlying stock’s move each session, then rebalances to that same target the next morning. Think of it as a rubber band that snaps back to the same tension every day. The compounding effect means that over weeks or months, the ETF’s cumulative return diverges badly from twice the underlying’s cumulative move, which is why NVDL gained 9.3% over the past year while NVIDIA itself gained 17.9%. These instruments are built for single-session tactical bets, not infrastructure budget cycles.
Based on reporting from SPCQ +21% and NVDL + 6% as SpaceX Earnings Reshuffle the AI Chip Trade, originally published 2026-08-05 12:30:00.
