Reviewed November 5, 2026. This analysis compares public Form 13F holdings for the quarters ended March 31 and June 30, 2026.
Did managers reduce their reported AMD and NVIDIA share holdings in Q2 2026, or did rising quarter-end values disguise lower share counts? Among the fixed 17-institution panel we track, the answer is clear for the managers that reported each stock in both quarters: aggregate AMD shares fell 16.0% across six continuous holders, while aggregate NVIDIA shares fell 6.0% across nine. Yet their reported values rose 139.9% and 7.8%, respectively. The divergence shows why a higher 13F value cannot, by itself, be described as additional buying.
The AMD result is also unusually concentrated. Whale Rock raised its reported AMD shares from 69,211 to 1,509,650 and represented 39.5% of the six continuous holders’ Q2 share total. Excluding Whale Rock, the other five managers’ combined AMD shares fell 48.4%. This is a comparison of public quarter-end positions—not trade dates, purchase cost, investment intent, total assets, or performance.
The answer in one table
| Continuous-holder comparison | AMD | NVIDIA |
|---|---|---|
| Managers reporting the stock in both quarters | 6 | 9 |
| Q1 → Q2 shares | 4,550,128 → 3,822,482 | 37,265,888 → 35,018,560 |
| Aggregate share change | −16.0% | −6.0% |
| Q1 → Q2 reported value | $925.6M → $2.221B | $6.499B → $7.007B |
| Aggregate reported-value change | +139.9% | +7.8% |
| Share increasers / decreasers | 2 / 4 | 3 / 6 |
| Eligible-equity weight increasers / decreasers | 5 / 1 | 3 / 6 |
Why share counts and reported values moved in opposite directions
A Form 13F reports both shares and quarter-end market value. For the same security, reported value can rise even when shares fall if the quarter-end price effect is large enough. Dividing the aggregated reported value by aggregated shares produces an implied quarter-end value per share of about $203.43 for AMD in Q1 and $580.91 in Q2. The corresponding NVIDIA amounts were about $174.40 and $200.09.
Those derived amounts are a consistency check on the filings, not an estimate of purchase price or investment return. They explain the mechanical direction of the value change: AMD’s implied quarter-end amount rose 185.6%, overwhelming the 16.0% decline in continuous-holder shares. NVIDIA’s implied amount rose 14.7%, more than offsetting a 6.0% share decline.
This distinction extends our earlier comparison of semiconductor weights across the full 17-institution panel. That study asked whether portfolio exposure rose. This one asks what happened underneath the weight: did continuous holders report more shares, or did quarter-end valuation do most of the work?
AMD: five of six weights rose, but four of six share counts fell
AMD provides the sharper counterexample to reading higher portfolio weight as automatic evidence of net buying. Five of the six continuous holders ended Q2 with a higher AMD weight inside our eligible public-equity denominator, but only Soros Fund Management and Whale Rock reported more shares. Appaloosa, ARK, Bridgewater, and Light Street reported fewer.
| AMD continuous holder | Q1 → Q2 shares | Share change | Reported value change | Eligible weight, Q1 → Q2 |
|---|---|---|---|---|
| Appaloosa | 221,400 → 197,500 | −10.8% | +154.7% | 0.80% → 1.64% |
| ARK | 2,712,854 → 1,416,443 | −47.8% | +49.1% | 4.35% → 5.39% |
| Bridgewater | 1,292,767 → 538,632 | −58.3% | +19.0% | 1.56% → 1.85% |
| Light Street | 219,251 → 109,500 | −50.1% | +42.6% | 8.97% → 7.56% |
| Soros Fund Management | 34,645 → 50,757 | +46.5% | +318.4% | 0.13% → 0.53% |
| Whale Rock | 69,211 → 1,509,650 | +2,081.2% | +6,128.7% | 0.18% → 7.04% |
Whale Rock is the decisive outlier. Its 1,509,650 Q2 AMD shares accounted for 39.5% of the continuous-holder total. Remove that one manager and combined AMD shares for the other five fell from 4,480,917 to 2,312,832, or 48.4%. The aggregate therefore does not describe a broad increase across the sample; it combines one very large expansion with four substantial reductions and one smaller increase.
The manager-level context is consistent with our earlier filing reviews of Whale Rock’s AMD expansion, ARK’s lower AMD share count, and Light Street’s lower AMD and NVIDIA positions. The new contribution here is putting those directions into one matched, reproducible comparison.
NVIDIA: reported value rose modestly while six of nine managers cut shares
NVIDIA’s divergence was smaller but still meaningful. The nine continuous holders reported 35.0 million shares in Q2, down 6.0% from 37.3 million in Q1. Reported value nevertheless rose 7.8% to $7.01 billion. Only Altimeter, Appaloosa, and ARK increased shares; the other six reported fewer shares.
| NVIDIA continuous holder | Q1 → Q2 shares | Share change | Reported value change | Eligible weight, Q1 → Q2 |
|---|---|---|---|---|
| Altimeter | 9,336,039 → 9,413,589 | +0.8% | +15.7% | 28.57% → 19.16% |
| Appaloosa | 1,471,500 → 1,525,000 | +3.6% | +18.9% | 4.58% → 4.37% |
| ARK | 1,035,426 → 1,381,247 | +33.4% | +53.0% | 1.42% → 1.81% |
| Bridgewater | 4,693,003 → 3,866,195 | −17.6% | −5.5% | 4.84% → 4.57% |
| Coatue | 6,331,620 → 6,055,197 | −4.4% | +9.7% | 3.80% → 2.49% |
| Light Street | 272,214 → 136,720 | −49.8% | −42.4% | 9.55% → 3.25% |
| Soros Fund Management | 1,073,206 → 1,064,635 | −0.8% | +13.8% | 3.52% → 3.80% |
| Tiger Global | 12,011,752 → 11,198,773 | −6.8% | +7.0% | 9.17% → 9.34% |
| Whale Rock | 1,041,128 → 377,204 | −63.8% | −58.4% | 2.35% → 0.61% |
The weight result is more mixed than the value result. NVIDIA’s eligible-equity weight rose at only three managers—ARK, Soros Fund Management, and Tiger Global—and fell at six. Altimeter illustrates why: it reported 0.8% more NVIDIA shares and 15.7% more value, yet the position’s eligible weight fell from 28.57% to 19.16% because the denominator and other positions also changed.
New, continuous and no-longer-reported positions are kept separate
We restricted the aggregate share-change calculations to managers that reported the same common-stock CUSIP in both quarters. Situational Awareness reported AMD and NVIDIA in Q1 but not Q2, and Third Point reported NVIDIA in Q1 but not Q2. Coatue and Tiger Global reported AMD in Q2 but not Q1. Those transitions are analytically important, but folding them into a same-holder percentage change would mix different questions.
“Not reported” is not synonymous with a complete sale. A position can fall below Form 13F reporting thresholds or be affected by the filing’s scope, and 13F does not show short positions or the rest of a manager’s assets. Likewise, a Q2-only report does not identify the transaction date or prove that every share was purchased during the quarter.
Methodology and what the comparison can—and cannot—show
- Panel: the same fixed 17 institutional filers used in AHNIM STUDIO’s Guru statistics. Politician disclosures are not included.
- Periods: quarter-end holdings as of March 31 and June 30, 2026. Q2 is our latest fully verified matched panel as of this review; the SEC’s 2026 filing calendar sets November 16 as the Q3 deadline.
- Security match and corporate actions: AMD common stock CUSIP 007903107 and NVIDIA common stock CUSIP 67066G104, with share units marked SH. The matched rows retained those CUSIPs in both quarters. AMD’s official split history shows its last split in August 2000, and NVIDIA’s latest split was effective in June 2024, so both Q1 and Q2 2026 observations are post-split; no additional adjustment was required for this comparison.
- Continuous holder: a manager reporting that same security in both quarters. Related reporting-account rows were aggregated within a manager before comparison.
- Eligible-equity denominator: public operating-company equity rows after excluding put/call positions, non-share units, identified warrants, a preferred depositary security, and an ETF. The cleaned panel contains 4,073 rows from 4,269 source rows.
- Reported value: the filing’s quarter-end market value. It is not purchase cost, capital deployed during the quarter, or profit.
- Limits: Form 13F does not disclose transaction dates, investment motives, complete portfolios, cost basis, short exposure, or performance. This analysis therefore describes public position changes, not trading strategy.
The next useful test comes after the complete Q3 2026 panel is available: whether AMD’s continuous-holder share total remains dominated by Whale Rock, and whether NVIDIA’s share reductions continue even if its reported value changes direction. Until then, the Q2 evidence supports a narrower conclusion: valuation effects lifted reported values, but the matched share counts declined for both companies.
For the broader archive of manager-level filings and statistical comparisons, see the AHNIM STUDIO Guru research hub.
Sources and original documents
SEC, Frequently Asked Questions About Form 13F (updated March 6, 2026) and the SEC’s 2026 13F filing calendar; AMD official stock-split history and AMD investor FAQ; NVIDIA investor FAQ and split history.
Q1 and Q2 2026 SEC information tables: Altimeter Q1 / Q2; Appaloosa Q1 / Q2; ARK Q1 / Q2; Bridgewater Q1 / Q2.
Coatue Q1 / Q2; Light Street Q1 / Q2; Soros Q1 / Q2; Tiger Global Q1 / Q2; Whale Rock Q1 / Q2.
Transition checks: Situational Awareness Q1 / Q2; Third Point Q1 / Q2. Calculations by AHNIM STUDIO from the original information-table rows; reviewed November 5, 2026.
This article is for informational and educational purposes only. It is not investment advice.
Leave a Reply