Burry’s 13F Signals a “Consensus AI Basket” Risk — He Put Size Behind NVIDIA, Palantir, and (Separately) Oracle

13F-HR holdings tables record option positions in “value” and “share-equivalent” terms. The investor economics are straightforward: put holders benefit when the underlying price declines. For an “AI basket” thesis, the practical market read is that the investor was willing to pay for downside exposure in the consensus winners—especially where valuation and expectations have expanded.

What this changes versus a 1999/2000-style “macro bubble only” call is the selectivity: it isn’t just “tech is expensive.” The document ties risk to specific names—NVIDIA (compute bottleneck expectations) and Palantir (enterprise AI/government-software rerating). That combination behaves like a bet on multiple compression across the AI stack rather than a single-industry slowdown story.

For NVIDIA, FY2025 revenue was $130.5B (vs. $60.9B in FY2024), and net income was $72.9B. For Palantir, FY2025 revenue was $4.48B (vs. $2.87B in FY2024) with net income $1.63B. The fundamental story isn’t collapse—it’s expectation-setting.

This is where the pair trade logic matters. NVIDIA sits at the hardware/accelerator chokepoint; Palantir is closer to software outcomes in enterprise/government. Buying puts on both simultaneously is consistent with a scenario where both the capex curve and the value-realization curve misalign—even if each company still reports earnings.

A full-stack AI drawdown typically propagates like this: semiconductor demand (accelerators) affects OEM and board-level pricing; hyperscalers and enterprises then govern the pace of training/inference deployments; software layers (platforms and decision tooling) rely on “effective usage” to convert spend into renewal and expansion.

So even though a 13F lists two names, the economic channel you should map is broader: accelerator utilization and networked deployments drive the software spend narrative. That’s why consensus “AI leaders” are often vulnerable together: when expectations reset, multiples compress across multiple tiers.

A practical checklist for investors:

• If you’re tracking “AI basket” downside, start with NVIDIA and Palantir because they are explicitly named in that 13F holdings table.
• If you want to include Oracle, you must source the linkage from a different verified disclosure (another 13F period, or a separate primary document).