Cathie Wood Just Dumped $134 Million of Broadcom and Bought Nvidia and SpaceX

During the Nasdaq 100’s worst session since March 2020, ARK sold 410,559 Broadcom shares, worth about $119.7 million at Thursday’s price, while buying 252,592 Nvidia shares and 393,774 SpaceX shares; ARK gave no reason for the trades. Broadcom fell 10.43% to $291.53, Nvidia dropped 9.26%, and SpaceX declined 9.05%; Broadcom later rose 1.92% in after-hours trading. The rotation contrasts with Broadcom’s fiscal Q3 revenue growth of 85.5% to $29.59 billion and Q4 guidance of $34.80 billion in revenue, including $21.70 billion from AI semiconductors. Nvidia reports November 17, ahead of Broadcom on December 10, providing an earlier test of AI demand.

Analysis

The rotation is a weak signal of relative conviction, not evidence that demand is shifting from custom silicon to GPUs: daily share counts omit execution prices, and ETF flows can reflect risk limits or liquidity needs. Economically, ASICs and GPUs can coexist across workloads; a durable AVGO impairment would require customer projects to be delayed or AI-capex budgets cut, not simply stronger GPU demand. The more useful near-term read-through is NVDA’s report: demand and customer commentary can reset expectations for the whole AI supply chain before AVGO reports. A strong NVDA result would not by itself validate AVGO’s custom-chip outlook, which needs customer/project-level evidence.

SPCX offers potential platform upside, but rapid growth from an undisclosed base does not establish material earnings power. Its AI ambitions may also raise capital and execution needs; widening credit protection costs are a warning that equity enthusiasm and financing risk can diverge. The political-risk narrative is a shared factor for all three names, but speculation without proposed rules is not yet a measurable earnings shock. Over 1–3 months, watch NVDA’s demand commentary, AVGO’s AI revenue versus guidance, and any concrete policy proposals. Over 6–18 months, the key question is whether hyperscalers sustain AI investment and diversify between in-house/custom chips and GPUs.

Key Decisions for Investors

• Do not trade the ARK print as a standalone signal. Treat it as a flow event unless subsequent filings or company-level demand data confirm a fundamental rotation.
• For a tactical AVGO long, wait for price stabilization around the cited 52-week low rather than averaging down automatically; exit or reassess on a close below that level, and cut the thesis if December AI revenue misses management guidance.
• Use NVDA’s November report as the first sector catalyst: if demand commentary weakens or the stock sells off despite strong results, reduce broad AI beta, including AVGO and SPCX; if demand is reinforced, assess AVGO separately for custom-chip project traction.
• Keep SPCX exposure smaller than a comparable liquid large-cap position until the scale, margins, funding requirements, and credit-spread move behind its AI business are verifiable. A sharp further widening in credit protection costs is a risk alert, not a reason to extrapolate headline revenue growth.