The 2nd-largest retail inflow day for NVDA landed one week before the Nov 17 earnings report. Retail pre-earnings buying tends to elevate price into the event — which can then produce a 'sell the news' response even on a strong print. The 2-3 trading days after Nov 17 will be more meaningful for trend assessment.

Burry's Broadcom-Anthropic structure question is worth watching because complex financing arrangements can obscure Anthropic's actual debt-to-asset ratio and cash flow. The key for IPO investors will be how these commitments are presented in Anthropic's S-1 — specifically when and under what conditions cash is actually received.

Three things to watch on Nov 17: 1) Blackwell shipments above Q3 estimate, 2) Data center revenue confirming $400B+ annual run rate, 3) Rubin 2027 guidance clarity. Two-out-of-three positive = consensus view is additional upside from here. This is also what's driving the record retail positioning ahead of the print.

NVDA Jan'29(E) FCF $452B — with P/E at 23× vs 10-yr avg 61×, if the FCF path holds, multiple normalization alone implies substantial upside. Key assumptions: agentic AI demand acceleration + Blackwell mass shipments. Nov 17 earnings is checkpoint #1.

xAI Colossus 2 tops 1.11M H100-equiv GPUs as world's #1 AI cluster — with 4 more clusters (420K GB300s) under construction. GB300 inference throughput is 20-30x H100, so the actual compute gap is far wider than the raw numbers suggest. This explains NVDA's Blackwell order pipeline.

Semi EV/Revenue: MU 4x, NVDA 9x, AMD 13x. NVDA at 9x reflects revenue denominator growing faster than EV. On margin basis NVDA remains premium tier — Blackwell-to-Rubin cycle transition could re-rate the multiple higher.

VANDA record #2 retail inflow one week before earnings. Retail is piling in ahead of the print. But this setup historically can trigger sell-the-news on earnings day even with a strong beat. Watch for short-term profit taking pressure regardless of print quality

Burry saying NVDA bear is shockingly hard to hold isn't just a personal complaint — it means AI infrastructure demand visibility is so strong that building a structural bear case is structurally difficult. 15% short interest is historically low. The cost of maintaining a short in an everyone-long environment is brutal

Burry tracker is unverified so don't take it at face value — but the core question is valid. Need to check Anthropic's S-1 directly to see if the Broadcom contract is a simple revenue agreement or if it has complex financial covenants attached. Reading the S-1 carefully before IPO is essential

NVDA 3-yr FCF heading for ~$1T: Jan'27 $190B, Jan'28 $327B, Jan'29 $452B. CAGR 58%. P/E at 23× vs 10-yr average 61× — if this FCF path holds, current price looks like a discount

xAI Colossus 2 just passed OpenAI Stargate to become #1 globally — and they're STILL building 4 more clusters adding 420K GB300s. At this point xAI has won the compute infrastructure race

MU at 4x, AMD at 13x EV/Revenue — 3x spread within the same sector. Market is brutally pricing AI exposure. NVDA at 9x just because the revenue base got so large it compresses the multiple

Agent AI using 4-10x more GPU than standard inference makes complete sense. A chatbot response is 1 inference. An agent planning, executing, checking, revising runs dozens. This explains why NVDA demand refuses to cool down

Meta $3B into NVDA + Oracle NVL72 Supercluster. Add SpaceX $120.5B + $40B financing + Meta + Oracle — demand is diversifying. Single-customer risk gone, total demand growing. Structure is getting better.