What caused Nvidia’s nasty reversal Friday? Look to the Fed for clues
$229 to $217. Twenty-four billion shares. Five hours. Gone. Just gone. I’m talking about what happened to the stock of Nvidia when I left the office at 11 a.m. ET Friday for a day off to spend time in the final throes of my garden. It touched $229.26 at the highs of the day, before closing at $217.55, down 4.6% for the day. I am glad I didn’t see it unfold. It might have been how sickening I felt back in July after watching Intel unravel in the wake of that great quarter that coincided with the unraveling of Situational Awareness, the AI-focused hedge fund run by boy wonder Leopold Aschenbrenner. If you remember that moment, we saw how quickly those downward moves could evolve. Intel jumped in extended trading the night it reported, only to sink the following day during regular trading. The sell-off in Nvidia came in the wake of a magnificent quarter , like Intel, where the move up was so powerful but not as powerful as the move down. Intel happened faster, but the pattern was ridiculously similar. One key difference: the selling hit Intel during the first session after reporting. With Nvidia, we saw a nearly 9% pop Thursday after reporting Wednesday night; it was Friday’s session when the sellers pounced, erasing more than half of Thursday’s gain. Oddly, the Situational shorts didn’t unravel until last week, when Salesforce had a celebratory gallop. It made me wonder how many people really believed in the “SaaSpocalypse” theory — the belief that software-as-a-service companies had no future in a world of vibe-coded applications and more “agentic workers” than humans. After all, we never got bad quarters from Workday or ServiceNow or Salesforce . They kept turning in decent quarters. I am convinced that Situational had more going for it than just the “long hardware” thesis. A betting man would say he made himself right with his shorts, whether they be Microsoft or Workday or all the others, including everything in the ill-fated IGV, the colloquial name for the iShares Expanded Tech-Software Sector ETF . I believe that Situational did something I had only seen when the Securities and Exchange Commission foolishly scrapped the uptick rule in 2007. The uptick rule forbade a short seller from knocking a stock down by endlessly selling it no matter what. Before that you had to wait for an uptick in trading, essentially a real buyer. Once that protection was abandoned, the stocks of many financials were easily obliterated. It isn’t “blamed” much, but the abandonment of the uptick rule, established to stop exactly what occurred in 2007-2008, was part of the swiftness with which institutions were felled. I think that Situational had a unique trading style that was the equivalent of getting short aggressively —and then loud — among fellow travelers. Consider each target wounded by crushing knockdowns and then killed by piggy-backers, irrespective of the quality of the quarter. It was possible, for example, to have ServiceNow report a very good quarter and then watch it fall rapidly after an initial feint higher (this happened in January and in April ). Situational was able to force journalists and commentators to come up with reasons why the stock had fallen so quickly. Sadly, this style of trading always works because the complex of those who opine on this kind of selling don’t know much about what they are talking about and totally take their cue from “the action.” We saw that same behavior in the first hour of extended trading Wednesday night, before Nvidia CFO Colette Kress spoke about the company’s surprise revenue growth forecast for fiscal 2028, which was insanely strong. I finally gave up trying to figure out what was wrong with software stocks that fell after good quarters and focused on what was right because you can’t foment what you know you don’t believe in. One had to wonder whether this excellent Salesforce quarter would have been met with similar selling, although the mid-August report that Silver Lake was eyeing a take-private deal for Workday might have assuaged buyers. I wonder, by the way, whether that that prospective deal is really dead. Perhaps another proud outfit like Thoma Bravo could have interest. You take Workday private, you demonstrate the health of the group and you change the entire narrative — so much so that private equity firms may see a path to bringing their software portfolio companies to the public markets. What narrative, however, should we ascribe the collapse of Nvidia’s stock on Friday? Again, the downside of a stock is rarely thought of as a casualty of the market itself. It’s more in the fault of themselves, to abridge a Shakespearean reference. We would say that Nvidia, it turns out, really is the province of circular reasoning. Or the 50% of customers away from the data center is shaky in capitalization. No matter what, it would never be blamed on Federal Reserve Chairman Kevin Warsh. To me that’s pretty ridiculous. I think Friday’s nasty decline coincided with Warsh’s harsh words during his first address at the Fed’s annual Jackson Hole symposium. The top U.S. central banker suggested that i nterest rates may need to move higher to tamp down inflationary pressures if current trends don’t improve. In response, the market odds of a rate hike at the Fed’s September meeting increased significantly, according to the CME Group’s FedWatch tool . Nvidia’s growth has now been tied to the politics of the data center — meaning the lack of political support to build among both parties, and the need to finance the buildout, which will grow more difficult if the Warsh-led Fed soon tightens. The Fed’s benchmark overnight lending rate influences borrowing costs across the economy. So, if it goes higher, the interest rates for other kinds of debt should follow. Nvidia’s giveback Friday was certainly not because the company didn’t report an amazing, if not breathtaking, quarter. I think that if Nvidia had a more substantial buyback program in place, it could have snapped up the entire sell mob. But it doesn’t yet subscribe to a Sandisk-like capital return program , although Nvidia is better than Micron, which has no buyback whatsoever for at least a few more months . Micron’s two-year ban on large-scale buybacks in order to secure CHIPS Act funding expires in December. I think Nvidia’s stock may be the first casualty of Warsh’s tough talk given the timing of when the selling started. At this stage of the AI buildout, borrowing money has become so important to new construction that one has to believe there lies the linkage. The new hurdle isn’t yet built in, but the common stock reacts quickly to any crimping, especially when it is up so much in one 24-hour period. I wish I could tell you when the slide might end. I figure that Monday we are filled with tales of woe: the end of financing, as if that were true, and fear that stocks are the loser in the upcoming bruising fight between Warsh and Treasury Secretary Scott Bessent. Why do I know this? Because I have been a journalist for as long as I have been a professional investor. The only two things you can be sure you will get out of the hedge-fund-reporter-columnist cadre are stories about how money will no longer be easy, sabotaging stock prices, and the obligatory story about how China is ahead of us no matter what. It’s such a shame that we get these worthless narratives so often. I know they can’t be defeated. When I would challenge these at thestreet.com, a site I built from the ground up, the justifiably rebellious editors would say, ‘Why don’t you just argue against them?” So, that’s what I would do. But the “facts” never change. You can’t defeat the press if it is solidly against a stock. It must be marvelous to be short a stock like Nvidia right now, as I thought the long knives would be sheathed by the quarter. That’s not going to happen when Warsh is on the warpath. What do you do about it? First, you need to separate the fundamentals from the bond-to-stock conundrum. You just can’t get the biggest stock in this market to trade on its own fundamentals when you see the algorithmic trading programs wash over stocks in endless fashion as we had Friday. Second, you can recall that Nvidia now sells at 14 times earnings estimates for fiscal year 2028 (it just reported second quarter fiscal 2027 results). Given that Nvidia’s earnings are accelerating, it is difficult to argue why this stock shouldn’t be the first to rally in a decent tape. The unassailability of the strength here must be obeyed or, at least, acknowledged. After that we need to see software cool off — the usual first victim of a sour bond market — and an Nvidia- led market can resume. Unfortunately, the only thing that can truly reverse the bond market’s fortunes is a definitive move down in oil, as we need to see an end to businesses passing through their higher input costs. It does seem that every company in any chain that has anything to do with diesel fuel is getting hammered and becomes a creature of oil. One of my biggest concerns, though, heightened by a new month, is the possibility of a Democratic sweep in the midterms — and I mean a full-out booting of the Republicans. We have a pro-business government right now, one which has to be acknowledged even if you hate President Donald Trump. We could have a chaotic new regime — chaotic in a different way — that would make any more favors for business strictly forbidden. A combination of executive office tariffs so horrendous for business that also do nothing to help the Treasury, and legislative handicaps to business that I don’t think can be vetoed, would spell a real hard time for any industrial, including tech industrial, hence Nvidia. I do wish I could be a little more positive. The midterm narrative will be so powerful — including the anti-data center story, sans those who cater to unions in a real battle royal — that it could drown out the positives. That’s why for the Club we might still be leaning to the sell side even as we have low-teens cash. I don’t think stocks like Johnson & Johnson and Wells Fargo will be the ones that hurt us. It will be tech. And tech means Nvidia, which means its next direction — whether that’s to new highs to reflect its book of business, or to new lows to reflect interest rates and elections — will determine much of the direction of the next move in the market. Rather bizarrely, that means Nvidia is in the hands of the oil market, because the oil market determines inflation and inflation determines Warsh’s thinking. If we have to face a fight with the Fed, we will lose. You can’t fight the Fed. And you can’t fight the tape, both of which will be compromised if this war doesn’t end and the price of diesel, far higher than oil, comes down so hard that price rollbacks will be obvious and Warsh changes direction on a dime. (Jim Cramer’s Charitable Trust is long NVDA, INTC, CRM, JNJ and WFC. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
