AMD and Broadcom gained about $0.57tn as Nvidia lost about $0.50tn, mostly before Huang quit

AMD and Broadcom had made 78% and 72% of their 6 to 12 November gains before Nvidia's CEO resigned. Earnings explain much of it, and no customer transfer is shown.

AMD and Broadcom were already rising on earnings and AI revenue, and the governance news added to it. That is the base case. AMD rose 23.1% and Broadcom 17.5% from the 6 November close to the 11 November close. Nvidia announced on 12 November that Jensen Huang had resigned as president, CEO and director, so most of the move came first.

Beyond that, investors may also have paid an option premium: extra value on a rival when buyers gain a credible alternative to Nvidia, ahead of any order. Prices alone cannot separate that from earnings. Broadcom reports fiscal fourth-quarter results on 9 December. AI semiconductor revenue at or below the $16.7bn of the prior quarter would weaken its earnings-backed rally, but would not by itself disprove a premium for supplier choice. Broadcom has guided to $21.7bn.

Seventy-odd percent was done before Thursday

Thursday's trading added a last leg. At about 18:30 UTC, AMD stood near $582.04, up 5.3% on the day, and Broadcom near $365.03, up 5.7%. Those are intraday prices, not closes. Including them, the gains since 6 November reach 29.6% for AMD and 24.3% for Broadcom.

Put differently, 23.1 divided by 29.6 is 78%, and 17.5 divided by 24.3 is 72%. That is the share of each rise already done by the 11 November close. The first leg ran from 6 to 9 November, per Yahoo Finance price history for AMD and Broadcom. AMD rose 13.3% and Broadcom 8.8% between the 6 and 9 November closes.

$0.57tn gained, $0.50tn lost, and no transfer

AMD gained (582.04 minus 449.04) times 1.663bn diluted shares, about $0.22tn. Broadcom gained (365.03 minus 293.69) times 4.887bn diluted shares, about $0.35tn. Together that is about $0.57tn. Nvidia lost (186.64 minus 165.79) times 24.15bn shares, the count implied by a $4.005tn market value at $165.87, about $0.50tn. All three are approximations, derived from Yahoo Finance prices and the share counts above, not market-cap quotes.

The catch is that similar size does not show money moving from one stock to the other. The benchmark is the earlier slide. Nvidia lost about $1.27tn between 6 October and 6 November, with no CEO event. Against 6 October closes, and using the 12 November intraday prices, Nvidia is down 30.7% and Broadcom 2.9%. Part of the rivals' gain is recovery.

The earnings case, and the part it cannot explain

AMD reported Data Center revenue of $8.5bn for the September quarter, up 97%. It credited EPYC processors and the launch of its MI450 GPUs. It listed Anthropic, Meta, Microsoft, OpenAI and Oracle Cloud Infrastructure as running workloads on its first Helios rack systems. Lisa Su said she expects growth to accelerate further in the fourth quarter.

Broadcom's AI semiconductor revenue in its fiscal third quarter was $16.7bn, up 221%. On 26 October, four financial partners announced a $63bn financing structure for accelerator purchases from Broadcom for Anthropic. An announced structure is not completed financing or funded purchases. It does show a planned financed route for at least one buyer to a non-Nvidia architecture, which is what an option needs.

The strongest case against a premium is the simple one. AMD's Data Center revenue rose 97% and Broadcom's AI revenue 221%, so both stocks had earnings to rise on, while Nvidia absorbed a governance discount. The dates support that: 78% and 72% of the move was done before 11 November, and both rivals remain below their 6 October levels, while Nvidia is down 30.7%.

Much of the gain is earnings momentum and recovery. Prices cannot separate a premium for choice from that, so it is the unproven part.

Nvidia's case, and the orders that have not moved

Bernstein's Stacy Rasgon cut Nvidia to Market Perform on 12 November. He cited how tightly Huang managed customers and suppliers. The same note, as relayed by Yahoo Finance, says the shares "screen extremely cheap". Nvidia's release quotes board chair Harvey Jones: "we expect the company's operations, customer commitments and product roadmap to continue without interruption."

Decrypt, in a piece carried by Yahoo Finance, put the limit plainly: "Those alternatives existed before Huang resigned, and Thursday's gains do not establish that Nvidia has lost customers." CTOL found no evidence of any customer moving orders away from Nvidia in company releases and the market coverage it read.

Wedbush, as relayed by Proactive, adds that supply constraints could slow how fast alternatives absorb demand. The lack of order transfers is compatible with an option premium, but the same facts are also consistent with an earnings-led rally.

A real transfer shows up in a dated series. Mercury Research data, as reported by Wccftech, has AMD's server CPU unit share at 25.1% in Q4 2024 (per an earlier Wccftech report) and 33.2% in Q1 2026. That is 8.1 points over five quarters, the product of customers buying over time, which a week's price move cannot show. These figures concern server CPUs, a different market from Nvidia's accelerators, so they show how adoption becomes measurable, not how this week's stocks are priced.

A financing unwind looks different too. Lucent's FY2002 annual report shows customer financing commitments of $8.1bn at 30 September 2000. Charges of about $1.8bn followed in fiscal 2001 and $765m of provisions in fiscal 2002. Lucent financed its buyers and booked charges against that financing. Nvidia's CoreWeave arrangement is, per Decrypt, an initial $6.3bn agreement to purchase unused capacity. Both can support customer spending, but Lucent recorded financing losses and CTOL found no comparable Nvidia loss or reported default. Customer financing can also strengthen Nvidia's moat.

Two earlier dates come first. Nvidia reports on 17 November, and Anthropic's shares are expected to start trading on 19 November, according to Yahoo Finance's live coverage. The date that matters most is 9 December. A Broadcom result near its $21.7bn guide would support that company's operating momentum, not prove the valuation of either rival or the source of its gains. Until then the rally is best read as earnings momentum plus a governance discount on Nvidia, with any price for choice unproven.