Nvidia 'Load Up by May 20' Trade: Holder Data Check

A TipRanks piece making the rounds this week carries a clean directive: 'Load Up Ahead of May 20,' Says Investor About Nvidia Stock. May 20 is the next NVIDIA earnings window. The argument is the now-familiar AI capex story, with the trade tucked in front of the print as a setup. Nine days out from that date, the more useful question for investors is which institutional holders are actually positioned to drive the post-print tape, and which ones are passive ballast that won't react to anything.

13F Insight tracks 6,041 institutional holders of NVDA across the most recent filings. Fifteen of the top twenty are flagged as active filers. But the dollar weight at the very top of the list is dominated by index money — and that's the disconnect every 'load up' trade has to reckon with before it gets sized.

The Passive Wall

The four largest reported holders are BlackRock ($362.80B), Vanguard Capital Management ($268.52B), State Street ($184.91B), and FMR LLC ($181.10B). Three of those four (BlackRock, Vanguard, State Street) are the index-fund complex. They hold NVDA at index weight because NVDA is the largest S&P 500 component by market cap. They will not 'load up' before May 20 — their mandate forbids it. Their position size is mechanical.

Adding the explicitly-classified passive overlay — Geode Capital Management ($109.45B, passive_index), Northern Trust ($47.33B, passive_index), and Charles Schwab Investment Management ($29.09B, passive_index) — puts roughly $1.2 trillion of NVDA dollar weight in the 'index mandate' bucket. That's the passive wall. Plus the market-maker rows — Jane Street Group ($37.89B, market_maker) and Susquehanna International ($36.11B, market_maker) — which represent option-hedge and inventory exposure, not directional bets. Reading any of these as 'institutional conviction' on the May 20 print is a category error. They are present in the ownership table because of structural mandates, not because anyone at those firms made a call about Nvidia's data center revenue trajectory.

Where the Active Conviction Actually Sits

Strip out the passives and market makers, and the real active layer in NVDA's top 20 has roughly nine names that matter:

• FMR LLC (Fidelity) — $181.10B. The largest active book by a wide margin. Fidelity Contrafund and the Magellan/Blue Chip suite have been long-conviction NVDA holders.
• JPMorgan Chase — $85.07B (active asset management + private bank book).
• Norges Bank — $62.24B (sovereign wealth, multi-year horizon).
• Morgan Stanley — $60.38B.
• Goldman Sachs — $41.83B.
• UBS AM — $36.53B.
• Bank of America — $34.91B.
• Capital Research Global Investors — $32.00B.
• Legal & General Group — $29.51B.

These nine, plus Citadel Advisors at $28.82B (multi-strategy with hedges), are the names whose post-print 13F moves will read as actual conviction shifts. The Citadel position deserves the usual asterisk — Citadel's gross long line items frequently sit on top of offsetting short or option positions, so the net directional read on any single quarter is incomplete from the 13F alone.

What the 'Load Up by May 20' Trade Is Actually Saying

The TipRanks formulation is essentially a binary bet that NVDA's data-center revenue and forward guidance on the May 20 print will beat the consensus narrative around AI-spend deceleration. The bull case lives in three numbers from the institutional record:

First, Capital Research Global Investors' $32.00B position is roughly the size that signals a multi-quarter active conviction view from the Capital Group complex — a firm that historically doesn't trim into earnings noise. If the post-May-20 13F shows Capital Research adding, that's a stronger signal than any analyst price target revision in the same window.

Second, Norges Bank's $62.24B sovereign wealth position is structurally sticky. Norges trims when the market-cap weight rises past its internal AI/tech concentration band, not when NVDA's earnings reaction is mixed. Their May 20 reaction will be muted by design.

Third, the absence of any active 13D/G filings on NVDA is itself the signal. Zero 13D activists, zero recent 13G crossings — there is no public catalyst hidden in the ownership tape, which means the May 20 print is the only catalyst with meaningful upside or downside vector right now. The 'load up' thesis is structurally a clean earnings play; it is not levered to an activism or M&A surprise.

The Insider Tape Is Quiet

One thing worth flagging: there are no recent insider transactions in our Form 4 tape on NVDA covering the past 90 days. Jensen Huang's previous large sales (covered in our Huang insider profile) were systematic 10b5-1 plan dispositions that concluded earlier this cycle. The absence of any pre-earnings insider sells is not a positive signal on its own — blackout windows around earnings discourage insider transactions in the run-up to a print — but it does mean the 'load up' trade isn't bumping into countervailing insider distribution.

Anchors to Watch After May 20

• 2026-05-20: NVIDIA Q1 FY27 earnings call (after-hours). Forward data-center guide and Blackwell ramp commentary are the primary moves.
• Q2 2026 13F deadline (mid-August 2026, 45 days after 2026-06-30 quarter-end): the first 13F window that will reveal whether Fidelity, Capital Research, and the active book added or trimmed in the May 20 reaction. The Q1 13Fs being filed this week will not yet show the post-print activity.
• Next 13D/G filings: any activist surprise would require a 13D, not a 13G. As of this writing none is pending. Track the smart-money signal feed for cross-stock confluence.

The 'load up ahead of May 20' line is a perfectly legitimate framing for a momentum trade, but it should be sized against an ownership picture in which most of the visible dollar weight will not react to anything in the print. The active conviction is real, but it sits in roughly ten names — not the four hundred billion at the top of the list. See NVDA's full 6,041-holder institutional map →

Investors building a framework around earnings-window holder reads will find the learn library useful for distinguishing passive from active filers without being misled by gross dollar weight.