CoreWeave April Sale: NVIDIA Still Owns 11.5%, Magnetar 14.9%

CoreWeave Chief Development Officer Brannin McBee filed a multi-line Form 4 on April 29 covering converted Class B shares sold into the open market on April 27. The transactions total roughly $9 million across a couple dozen execution lines, priced between $105 and $112. For a post-IPO insider on a name still digesting its listing, this is unremarkable on its face — the kind of cadence that a 10b5-1 program throws off on schedule.

What's worth reading is two filing types over. CoreWeave's Schedule 13G docket over the last three months tells a story that almost no other recently public US company can match: the company's chip supplier owns 11.5% of it, a credit-and-PIPE specialist holds 14.9%, and Goldman Sachs filed a fresh 8.1% stake last week.

The 13G stack post-IPO

The combined declared block from the four largest holders (NVIDIA, Magnetar, Goldman, Vanguard) totals approximately 41.7% of shares outstanding. Strip out Vanguard as passive and the remaining three holders — all of them with concentrated, non-index motivations — account for 34.5% of the float.

That is an unusual concentration for any large-cap, and especially unusual for a company that listed within the last twelve months. It also matters for how reading McBee's $9M Form 4 actually fits the picture.

The NVIDIA strategic stake at 11.5%

The single most distinctive line on CoreWeave's cap table is NVIDIA holding 11.5% of the company via its January 26, 2026 Schedule 13G/A. NVIDIA is CoreWeave's primary GPU supplier and the source of the H100/H200/Blackwell inventory that CoreWeave's hyperscaler-grade rental business is built on. NVIDIA's stake was originally acquired during a pre-IPO funding round and disclosed in the IPO prospectus; the 13G amendment formalized it as a post-listing reportable holding.

There is no comparable cap-table position in US listed equities right now: the chip supplier holding a 10%-plus declared block in its largest pure-play GPU rental customer is a single-counterparty concentration on both sides of the trade. From a 13F reader's perspective, the relevant fact is that NVIDIA's stake is not for sale in any meaningful sense over the near term — it is a strategic position, not a portfolio bet. Treat it as effectively locked free-float, the same way you would treat a founder-controlled Class B block on a tech name.

Magnetar at 14.9%

Magnetar Financial's 14.9% position (per its May 6 13G/A) is the other structural anomaly. Magnetar is a credit-and-volatility specialist whose CoreWeave position likely originated through pre-IPO PIPE or convertible exposure that converted into common shares at listing. At 67.97 million shares, this is the single largest declared holder on the docket — larger than the NVIDIA strategic stake by share count.

Magnetar's holding behaves differently from NVIDIA's. Where the NVIDIA position is functionally indefinite, Magnetar's exposure is more likely to unwind over time as the original credit structure pays down or the manager rebalances the converted equity into other ideas. The 13G/A amendment is the filing to watch for forward signal: if Magnetar's stake drops sequentially into Q2 and Q3 2026 13G/A filings, the disclosed-block selling pressure is real.

The Goldman fresh 13G at 8.10%

The May 7 SC 13G from Goldman Sachs is more interesting than a typical bulge-bracket 13G appearance. At 8.1% with 33.77 million shares, the position size suggests Goldman is reporting on behalf of a combined book that includes underwriting-related inventory, prime brokerage exposure, and managed-account positions. The 13G initial filing (not an amendment) implies the threshold crossing is recent — i.e., Goldman either built the position quickly through Q1 2026 or aggregated previously sub-threshold holdings under a single reporting umbrella.

Either way, Goldman at 8.1% adds to the structural concentration on the cap table. Between NVIDIA, Magnetar, and Goldman alone, three counterparty-level holders account for 34.5% of the company.

What McBee's $9M actually means in context

Back to the April 27 Form 4. McBee, as Chief Development Officer, holds 6,991,660 shares of derivative-track exposure on the Table II line after the April conversions and 0 Class A on Table I after the same-day sale — the standard convert-and-drip mechanic visible across other founder-led tech names (see also Roku's April pattern and Cloudflare's same-week filing).

What separates CoreWeave from those comparisons is the structural concentration on the cap table. McBee's 16.5M-share personal 13G/A position at 4.1% is large in absolute terms but small relative to the NVIDIA, Magnetar, and Goldman blocks above him. A founder-equivalent insider sale of $9M is roughly 0.5% of his declared 13G holding — and a vanishingly small fraction of the total declared institutional book.

For a stock with this much structural concentration, the relevant supply-and-demand questions are: what does Magnetar do at the next 13G/A; does NVIDIA ever trim the strategic stake; and how does Goldman's aggregated 8.1% book behave through 2026. None of those questions are answered by a $9M insider Form 4.

What to watch

• Magnetar's next 13G/A: a step-down from 14.9% would be the most directional signal on the docket. Magnetar is the holder most likely to actively trim.
• NVIDIA's next 13G/A (if any): the strategic-stake mechanic means amendments are rare. Any reduction from 11.5% would be a meaningful change of posture and worth treating as a sector-level signal, not just a stock-level one.
• Q2 2026 13Fs (due August 14): first quarterly read on whether active managers built CoreWeave as a position post-listing. Capital Research, FMR, Wellington, and the GPU-thesis specialists are the names to monitor.
• Goldman's 13G/A: whether the 8.10% block grows or normalizes over the next 90 days will tell you whether the initial 13G was a structural one-time threshold crossing or an active build.

See CoreWeave's full institutional ownership breakdown for the holder-level detail, and Brannin McBee's transaction history for the cumulative shape of insider activity since the IPO.