🧠 Food for thought

Implications, context, and why it matters.

Super Micro guidance implies near-doubling while peers hit by export controls

• Super Micro guides Q2 revenue of $10–11 billion, a 100–120% jump from Q1’s $5 billion, while NVIDIA could not ship $2.5 billion of H20 in Q1 after the April 9 U.S. export controls on advanced chips to China 12. The H20 is an NVIDIA GPU configured to comply with those limits.
• Q1 operating cash outflow was $918 million due to higher accounts receivable (cash not yet collected from customers) plus inventory build (components and finished systems staged for delivery) to support Q2 1. Non-GAAP gross margin was 9.5% versus 9.6% in Q4; Non-GAAP adjusts items from Generally Accepted Accounting Principles (GAAP). Management cited product mix plus ramp-related costs (temporary inefficiencies with production scale-up) as near-term headwinds.
• NVIDIA’s Blackwell NVL72 is shipping at nearly 1,000 racks (server cabinets) per week across major hyperscalers (the largest cloud platforms) 2. Super Micro must ship about 2–2.2x Q1 volumes in roughly 90 days while carrying $4.8 billion in debt against $4.2 billion in cash 12. The NVL72 is NVIDIA’s rack-scale, NVLink-connected Blackwell system for high-density AI training and inference.

Server integrators gain as AI factories and networking scale

• NVIDIA counts nearly 100 AI factories this quarter, double year over year, with average GPUs per site also doubling 2. AI factories are large, purpose-built data centers that train and serve AI models. This opens profit potential for server integrators (companies that design and assemble customized servers) that provide advanced thermal, power, or networking beyond basic racks.