Good Thursday —
SpaceX filed confidentially with the SEC on April 1, targeting a June listing at a $1.75 trillion valuation and up to $75 billion in proceeds, which would make it the largest IPO in history.
The framing in most coverage treats this as a space story. It's not, really. Following February's merger, SpaceX now owns xAI (Grok) and X, making it a vertically integrated space-AI conglomerate going public on an AI-forward roadshow narrative. The thesis isn't that capital rotates away from AI infrastructure names. SpaceX is positioning as one of them.
The structure of this deal is a cross-subsidy argument dressed as a valuation argument. The Starlink side cleared roughly $8 billion in profit last year on $15–16 billion in revenue. xAI is burning approximately $1 billion per month. At $16 billion in 2025 revenue, the $1.75 trillion target implies over 100x revenue. The roadshow pitch is essentially: Starlink's recurring satellite cash flow can carry xAI's model ambitions long enough for orbital data centers to become economically real. That's a specific and unproven bet. I don't think it's obviously wrong. I do think it's a much shakier foundation than the headline number implies.
One technical detail that doesn't get enough attention: Nasdaq recently changed rules that could allow SpaceX to join the Nasdaq 100 within 15 days of listing, triggering forced buying from index-tracking funds. That's a meaningful first-day bid with nothing to do with fundamentals. It inflates the opening dynamics and probably the valuation anchor for a while after.
Signals
Intel consolidates Fab 34 ownership. CPU margins and fab utilization now fully captured; shifts Intel's cash flow and capex flexibility.
NVIDIA and Marvell expand NVLink ecosystem for semi-custom silicon. Builders gain design flexibility; reduces NVIDIA lock-in risk.
Citi forecasts DRAM ASP +190% YoY, NAND +172% YoY for 2026. Server OEM margins compress; memory cost input surges flow through supply chain.
Alibaba Qwen3.6-Plus: 1M context, agentic workflows, Claude API compat. Model release; no unit economics or competitive shift disclosed.
Regime Snapshot
Compute (CRS): 66 — scarcity. Demand and supply roughly in balance. Watch for directional shifts.
Memory (MRS): 67 — Tight. Demand and supply roughly in balance. Watch for directional shifts.
Narratives Moving Today
This briefing is the summary. The full Tessara terminal tracks all 81 supply chain nodes in real time — live regime scores, stress maps, narrative momentum, and the data behind every signal above.