Filers Agree About Nvidia and Argue About the Firms Paying It
13F filers moved near-unanimously into Nvidia, Apple and Tesla last window while splitting almost evenly on Microsoft and Amazon — and no disclosure channel corroborates either side.
Sterling scan summary
13F filers are near-unanimous on Nvidia and split on the hyperscalers paying it.
Why Sterling flagged it
- 13F rotation flagged: 14 names with at least 6 filers moving the same direction over a 45-day window; the top six by disclosed value include NVDA at 12 adders against 2 reducers, AAPL at 10 against 1, TSLA at 7 against 1.
- The contested names in that same sample are the three hyperscalers — MSFT 9 adders/6 reducers, AMZN 9/7, GOOGL 9/4 — a one-directional share of 60%, 56% and 69% against 86–91% for the other three.
- Dollar weight is not where filer counts are: AAPL shows $90.2bn of disclosed movement on 11 moving filers, roughly 2.9x NVDA's $30.9bn on 14.
- News pressure flagged: 150 tickers cleared 6+ articles in 48 hours, with NVDA at 139 articles — more than double the next operating company (GOOG, 62), ahead of MSFT (53) and MU (40).
- No insider buy cluster, no insider sell cluster and no congressional cluster qualified — no disclosure channel corroborates the 13F asymmetry in either direction.
How Sterling got here
- Observation
The 13F rotation check flagged 14 qualifying names over a 45-day window. Within the top six by disclosed dollar value, one-directional share ranges from 91% (AAPL, 10 adders/1 reducer) to 56% (AMZN, 9/7). Nvidia sits at 86% (12/2), Tesla 88% (7/1), Alphabet 69% (9/4), Microsoft 60% (9/6).
- Cross-check
The news pressure check, an independent source, flagged 150 tickers at 6+ articles in 48 hours. Its leader is the same name filers are most one-directional on: NVDA at 139 articles, against GOOG 62, MSFT 53 and MU 40. Attention and filer conviction are concentrated on the same node of the complex.
- Independent confirmation
The disclosure channels returned nothing. Insider buy clustering, insider sell clustering and congressional clustering all came back clear with a qualifying count of zero. Sector breadth was not evaluated for this historical scan. So the configuration rests on two sources, not five — a complication, not a contradiction.
- Inference
Taken together, this would be consistent with institutional filers pricing the AI supply chain and the AI spending base as separate risks rather than one. Disclosure data shows positioning, not intent, and 13F data is lagged by construction: it describes where filers were at quarter-end, not where they are now. The reading is directional, not timed.
- Status
What is not corroborated: no insider or congressional activity supports either leg, dollar concentration sits in Apple ($90.2bn) rather than in the names driving the narrative, and the sample shown is the top six of 14 qualifying names — the remaining eight are not visible in this bundle and could point either way.
What data Sterling analyzed
Six deterministic checks ran against the 2026-06-02 bundle; two flagged.
| Check | Sources | Result | Detail |
|---|---|---|---|
| 13F institutional rotation | 13F filings, 45-day window | FLAGGED | 14 names with 6+ filers moving the same direction; top six by value show a 56%–91% spread in one-directional share. |
| News pressure concentration | Scored news feed, 48-hour window | FLAGGED | 150 tickers at 6+ articles; NVDA leads at 139, more than double the next operating company. |
| Insider buy clustering | Form 4 filings, 7-day window | CLEAR | No qualifying buy cluster. |
| Insider sell clustering | Form 4 filings, 7-day window | CLEAR | No qualifying sell cluster. |
| Congressional trade clustering | STOCK Act disclosures, 14-day window | CLEAR | No qualifying congressional cluster. |
| Unusual sector move | Daily sector breadth (live sessions only) | CLEAR | Not evaluated for historical scans. |
What would confirm
- Broadcom, June 3 (consensus EPS $2.40) — an AI-supply print that lifts supplier attention while hyperscaler commentary turns to capex discipline would extend the supply/spend distinction beyond positioning data into fundamentals.
- Next 13F window — hyperscaler reducer counts rising above the current 6 (MSFT) and 7 (AMZN) while Nvidia's holds at or below 2 would widen the gap that filer-count base rates cannot explain.
- Attention persistence — Nvidia retaining a 2x-plus article lead over the next operating company in a subsequent 48-hour window would keep the two flagged sources convergent rather than coincident.
- Breadth beyond the sample — the same supply-versus-spend sort appearing across more of the 14 qualifying names, not just the six visible here.
What would invalidate
- Convergence — hyperscaler one-directional share moving back toward the 86–91% band of Nvidia, Apple and Tesla in the next window, which would collapse the distinction entirely.
- Base-rate confirmation — the next window's most two-sided names again being simply the names with the most moving filers, regardless of business model, would favor the mechanical explanation over this one.
- Macro override, June 5 payrolls — consensus looks for 85k against 179k prior with unemployment at 4.3%; a print that repriced the whole complex together would submerge any within-group asymmetry in beta.
- Contradicting disclosure — a qualifying insider or congressional buy cluster appearing in the contested hyperscaler names, which would cut against the reading that filers are backing away from the spend side.