11/17/2026 | Press release | Distributed by Public on 11/17/2026 03:22
Today's big story: Could Nvidia Stock Fall 50%? Follow The Cash
This morning you probably asked Google something and got a written answer instead of ten blue links, or let Gemini draft a reply in Gmail. Google says its Gemini models are trained on its own chips, called TPUs, not on Nvidia's. Anthropic's IPO filing names Broadcom as Google's partner on them. Anthropic, the maker of Claude, trains on them too. So here is the question worth asking before tonight's Nvidia report: if Nvidia stumbles, does Google? And is there a stronger AI stack that doesn't run through Nvidia, or only a more expensive one? We'll come back to the cost. It's the part most coverage this week will skip.
Why now: Nvidia announced on Nov 12 that Jensen Huang had resigned as CEO, with no reason given, four days after its board opened an investigation into reports he was the target of a foreign intelligence effort. Nvidia reports third-quarter results for fiscal 2027 after the close today, its first report without him, and House committees have asked Huang for a closed-door interview on Dec 1. Since the Nov 6 close, Nvidia is down 16.3%, Broadcom up 13.5% and Alphabet down 0.4%, against a 0.6% gain for the S&P 500 fund (SPY). The market has already voted that Nvidia's problem is not Google's. Monday showed the other side: as tech led a broad fall, Alphabet dropped 4.1%, as much as Nvidia.
That vote is a reminder that the stock everyone watches is rarely the whole story. A portfolio that never rested on one is the Trefis High Quality (HQ) Portfolio : quality stocks in equal weight, ahead of its benchmark since launch.
Does An Nvidia Crisis Slow Google?
Not the kind Nvidia is having. We pulled every day since January 2015 when Nvidia fell 5% or more and set two kinds side by side: days it fell alone, with the S&P 500 fund down less than 0.5%, and days it fell with the whole market.
| Nvidia's bad day | Days | Alphabet | Broadcom |
|---|---|---|---|
| Nvidia fell alone | 18 | −0.4% | −1.2% |
| Nvidia fell with the market | 64 | −3.7% (fell 64 times) | −4.4% (fell 63 times) |
Median same-day move. Nvidia one-day falls of 5%+, Jan 2015 to Nov 16, 2026; "alone" = S&P 500 fund down less than 0.5% that day; "with the market" = down 1.5% or more. Source: Trefis, prices through Nov 16.
Two different animals. When the trouble was Nvidia's alone, Alphabet slipped a median 0.4%. When Nvidia fell with the market, Alphabet fell every single time, 64 out of 64, and by more than the S&P 500 fund's median 2.9%. Over the last 12 months, Alphabet's daily moves tracked Nvidia's less closely than the S&P 500's did: a score of 0.22 on a scale where 1 is lockstep, against 0.50 for Broadcom and 0.63 for the S&P 500 fund. So a Huang-style crisis doesn't slow Google. A market-wide sell-off slows everyone, Broadcom more than Alphabet. What a founder's exit has meant for other companies is in What Happens To Nvidia Stock Without Jensen Huang?
Is Google's Own AI Stack Already Strong?
Google Cloud rents out both Google's own chips and Nvidia's, so its growth is not a pure TPU number. Still, set the three AI businesses side by side and the Google stack is growing as fast as Nvidia itself: Google Cloud's revenue rose 103% in its latest quarter, Nvidia's 106%.
| Business | Latest quarter | Revenue | vs a year before |
|---|---|---|---|
| Nvidia, whole company | May-Jul 2026 | $96.2B | +106% |
| Google Cloud | Jul-Sep 2026 | $30.7B | +103% |
| Broadcom chips | May-Jul 2026 | $20.8B | +127% |
Latest reported quarter vs the same quarter a year earlier; Broadcom chips = its semiconductor solutions segment. Source: Trefis (Morningstar statements); Alphabet and Broadcom filings for segments.
Two more numbers say this is a business, not a science project. Google Cloud lost $1.9 billion in 2022, as Alphabet now reports it; last quarter it earned 36.2% of its revenue as operating profit, up from 23.7% a year earlier. And Broadcom's AI chip revenue reached $16.7 billion in its latest quarter, up 221% in a year, by its own filing. Broadcom doesn't say how much of that is Google. Anthropic's IPO filing commits it to spend at least $121.3 billion with Google Cloud through July 2033, but does not say how much of its training runs on TPUs.
What Does Building Without Nvidia Cost?
Here is the part the chip headlines skip. Not buying from Nvidia doesn't mean not paying. Alphabet pays for the chips, the buildings and the power itself, and the bill has outgrown its cash.
| Period | Capital spending | Cash from operations | Share spent |
|---|---|---|---|
| 2019 | $23.5B | $54.5B | 43% |
| 2023 | $32.3B | $101.7B | 32% |
| 2025 | $91.4B | $164.7B | 56% |
| Year to Sep 2026 | $159.8B | $179.6B | 89% |
| Jul-Sep 2026 | $51.4B | $42.3B | 121% |
Alphabet. "Year to Sep 2026" = the four quarters Oct 2025 to Sep 2026. Source: Trefis (Morningstar cash-flow statements).
Free cash flow, what's left after that spending, was negative in each of the last two quarters: minus $5.9 billion, then minus $9.1 billion. To close the gap, Alphabet raised $49.6 billion in June by selling stock and preferred shares, and borrowed about $49 billion in new notes over the second and third quarters. Its capital spending over the last year, $159.8 billion, came to about half of everything Nvidia sold in its last four reported quarters, $303.0 billion.
Investors have noticed. They now pay 7.4 times sales for Alphabet, 15% less than a year ago, while those sales grew 22%. Nvidia's multiple fell 56% and Broadcom's 34% over the same year. So, can Google build stronger AI without Nvidia's chips? On the engineering, the numbers say it already does. The constraint isn't Jensen Huang. It's Alphabet's cash, and tonight's report won't change that.
How Low Can They Go?
| Fall | Alphabet | Broadcom | Nvidia |
|---|---|---|---|
| 2007-09 | −65% | listed Aug 2009 | −85% |
| 2020 | −31% | −48% | −38% |
| 2022 | −44% | −35% | −63% |
| 2025 | −30% | −40% | −37% |
| Now, from 52-week high | −30% | −31% | −35% |
Peak to trough within each period; "now" = Nov 16 close. Source: Trefis, prices through Nov 16.
All three are already 30% to 35% below their highs, while the S&P 500 fund sits 5% below its own. The whole stack has been marked down, not just Nvidia. And the record shows how long a deep fall can last: in 2022 Alphabet dropped 44% and needed 15 months after the bottom to get back to its old high; Nvidia dropped 63% and needed 7.
What Is The Other Possibility?
That going it alone costs Google more than it saves. Google still buys Nvidia chips for cloud customers who ask for them, and Anthropic also trains on Amazon's chips and Nvidia's. If tonight's report shows Nvidia's demand still climbing and a smooth hand-over, the urgency behind custom chips fades a little. If it shows AI buyers slowing, the first table says Alphabet and Broadcom fall with Nvidia, whichever chips they use. What the report itself may reveal is in Could Nvidia Stock Fall 50%? Follow The Cash.
How Do You Position Whichever Way Tonight Goes?
What Should You Watch?
The question that outlasts tonight's report: how long can Alphabet spend more than its business brings in before investors push back?
Nobody can answer that in advance, so it helps if your plan doesn't ride on any single company's big bet but on many quality businesses, held in equal slices.
Why Own One Leader When You Can Own 30?
If you like finding quality companies like Alphabet, the Trefis High Quality (HQ) Portfolio holds 30 of them in equal weight, and has returned more than 105% since its September 2020 launch. You're welcome to look inside.
For education, not a recommendation to buy or sell.
For questions about this article or underlying research, email [email protected].