Nvidia Is on a 7-Day Losing Streak — Right Before Earnings

Okay, so this one actually surprised me.

Seven straight days. That's how long Nvidia has been dropping heading into Wednesday's earnings report — its most closely watched in probably two years. Seven days of investors looking at one of the most dominant companies in the history of the semiconductor industry and saying, you know what, I'm going to sell this right now.

Which is... a choice.

Now, before you read too much into it as a bearish signal, there's a very specific pattern that plays out around mega-cap tech earnings — especially one with this much anticipation baked in. But the context around this particular selloff is more complicated than just "traders taking profits." There are some real structural things happening underneath that are worth paying attention to.

Let me walk through what's actually going on.

The Setup: A Stock That's Carried Enormous Expectations

For the past two-plus years, Nvidia has basically been the face of the AI trade. Every quarter, the company beat estimates by wide margins. Revenue grew at a pace that genuinely looked made up. Data centers couldn't get enough H100s, then B200s. The stock became a proxy for the entire AI buildout thesis — if you believed AI was going to reshape the economy, you owned Nvidia.

And that positioning is exactly the problem going into this print.

When a stock goes from roughly $100 to several hundred dollars in under two years, the people who own it aren't just passive index investors anymore. You've got hedge funds that are massively concentrated, active managers who benchmarked against it, and options traders who've been whipsawed by it every single earnings cycle. A lot of those holders are nervous. And "nervous" translates to selling pressure before a binary event.

The seven-day losing streak heading into Wednesday isn't random noise. It's the market quietly pricing in the possibility that this could be one of those "beat and retreat" quarters — where the numbers are technically great but the guidance or the margin story or some random commentary about Blackwell chip yields spooks everyone into selling the news.

We've watched this movie before with Nvidia. Several times.

What the Actual Numbers Need to Look Like

Here's what I actually think about this — the bar for Nvidia is genuinely insane. Not in a hyperbolic way. Literally: the expectations the market has priced in require a level of execution that almost no company in history has had to consistently clear.

Analysts heading into this report are expecting something in the neighborhood of $43–45 billion in quarterly revenue, which would represent roughly 110–115% growth year-over-year. Data center revenue — which is basically the whole business now — needs to keep accelerating, not just hold steady. And gross margins, which briefly wobbled during the Blackwell architecture transition, need to show that they've stabilized or improved.

If all of that lands cleanly? The stock probably bounces.

But if data center revenue comes in at $38 billion instead of $43 billion — still an astronomical number by any normal measure — this stock gets punished. Because the market isn't valuing Nvidia on what it is right now. It's valuing it on what it's assumed to become. Any crack in that story, even a small one, triggers a repricing.

That's the uncomfortable reality of owning a stock at this kind of valuation multiple. You don't get credit for "pretty good." You only get credit for "spectacular, and here's why next quarter is even better."

The Broadcom-Marvell Factor — and Why This Time Feels Different

Now here's where it gets interesting.

The other piece of this week's chip story — and it's directly connected to why Nvidia's stock has been struggling — is that Alphabet quietly announced it's adding Marvell Technologies as a chip supplier, alongside Broadcom. Which sent Broadcom's stock down more than 10% in two weeks.

Sit with that for a second. Broadcom, one of the most important companies in the custom chip (ASIC) space, lost 10% of its market cap because one customer decided to diversify its supplier base. That's how fragile the "winner-take-all" narrative has become.

And the same logic applies to Nvidia, just at a different layer of the stack. Hyperscalers — Google, Microsoft, Meta, Amazon — have been openly public about their plans to design more of their own AI chips in-house. Not to fully replace Nvidia. But to reduce their dependency on a single supplier that, frankly, has enormous pricing power over them right now. The Marvell-Alphabet deal is a concrete data point that this diversification is actually happening, not just being talked about.

Does this crater Nvidia's business tomorrow? No. Absolutely not. Demand for GPU compute is still outpacing supply by a wide margin, and nothing Marvell or Broadcom makes can run general-purpose AI training workloads the way Nvidia's CUDA ecosystem can.

But the market is slowly adjusting its "Nvidia gets 100% of AI infrastructure spending forever" assumption. And that adjustment, spread across seven trading sessions of selling, is what you're watching right now.

Jackson Hole Is Sitting Right Behind This

I'd be doing you a disservice if I didn't mention what else is happening this week — because it's creating a weird double-pressure situation.

The Fed's annual Jackson Hole summit is underway, and the whole market is trying to parse what Chair Powell and company are signaling about the path for rates from here. We've had some pretty hawkish commentary from Fed officials in recent weeks, and any suggestion that the Fed is going to keep rates "higher for longer" into 2027 is a direct headwind for high-multiple growth stocks.

Nvidia trades at a forward P/E that makes traditional value investors cry a little. High rates compress valuations on stocks like this in a very mechanical, discounted-cash-flow kind of way. So you've got macro pressure from one direction and stock-specific uncertainty from the other, and they're both landing in the same week.

It's a lot.

And look, I could be wrong here, but I think the Jackson Hole noise is probably less important to Nvidia's post-earnings move than the earnings themselves. If the numbers are good, the stock rallies through macro concerns. Markets have demonstrated that repeatedly. But right now, in the uncertainty before the print, the macro backdrop is making cautious holders feel more justified in reducing exposure.

The Debt-Fueled AI Build-Out Angle

There's a bigger picture conversation happening this week too — MarketWatch ran a piece arguing that the entire AI infrastructure buildout might already be "too big to fail," borrowing language straight from the 2008 playbook. The idea is that the Fed's corporate credit facilities, which were initially a pandemic tool, have essentially capped the downside on massive AI infrastructure debt, because the hyperscalers and cloud providers taking on this debt are too systemically important to let fail.

That's a genuinely unsettling framing. And honestly? It tracks.

Microsoft, Google, Meta, and Amazon have collectively committed to spending something in the range of $300+ billion on AI-related capital expenditure in 2026 alone. Most of that gets financed through some mix of cash flow and debt markets. If the AI revenue projections that justify those investments turn out to be too optimistic — even by 20-30% — you've got a wave of corporate balance sheet pain concentrated in exactly the companies that underpin the broader market.

Nvidia is both the beneficiary of that spend and exposed to a slowdown in it. If hyperscalers pull back on data center orders to protect their own balance sheets, the first call they cancel is the next Nvidia GPU order. How much of this shows up in Wednesday's guidance is arguably more important than the revenue number itself.

What Happened Last Time Nvidia Dropped Before Earnings

Here's something that's easy to forget in the middle of a pullback: pre-earnings selloffs in Nvidia have not historically predicted bad results. If anything, the pattern over the last several cycles has been that the stock gets sold into earnings as traders de-risk, the actual results come in strong, and then it rips back.

The counter-example — the one where the selloff was a warning sign — was in early 2024 when concerns about Blackwell timing first surfaced. The stock held up into earnings, beat the top line, and then sold off anyway on margin concerns. That's the scenario traders are most scared of repeating.

My honest take: Wednesday night's report will tell us a lot about whether the broader AI trade has legs into 2027 or whether we're entering a period of more modest, "show me the revenue" skepticism. A blowout print with strong guidance would be a genuine confidence boost for the whole sector. A miss, or even a technically okay quarter with softer forward guidance, could be the moment the market finally starts repricing the AI trade more seriously.

Dick's Sporting Goods Just Added a Cautionary Data Point

On a totally different note — Dick's Sporting Goods tanked today after slashing its full-year outlook, citing "challenging conditions." That's worth a sentence or two because it's a real-world signal that the consumer spending story is getting bumpier.

The AI trade and the hyperscaler buildout exist somewhat insulated from the regular consumer economy. But they don't exist in a vacuum. If consumer spending is weakening and the macro backdrop is uncertain, that's another reason investors get more risk-averse about holding concentrated positions in high-multiple stocks — like, say, Nvidia — through binary events.

What to Actually Watch on Wednesday Night

So here's the part that actually matters. When Nvidia reports after the bell on Wednesday, these are the specific things that will move the stock:

Data center revenue — The consensus is somewhere around $43 billion. Anything in that range or above probably gets a positive reaction. Anything below $40 billion is going to hurt.

Gross margins — Nvidia's gross margins were briefly compressing during the Blackwell ramp. The market wants to see them stabilize in the 73–75% range. If they're sliding toward 70%, that's a problem.

Guidance for Q3 — This is often more important than the actual quarter. A conservative guide is how Nvidia has historically set up a beat. A guide that's outright disappointing would be rare but devastating.

Commentary on hyperscaler demand — Listen for whether management is seeing any signs of slower order velocity from the big cloud providers. One careful word choice here can move the stock 5% in after-hours.

The seven-day losing streak isn't a verdict. It's a question. Wednesday night is the answer.

What This Means If You Own NVDA (or Are Thinking About It)

• If you already own it and you're a long-term holder: You've probably been through enough Nvidia earnings cycles to know that the pre-report volatility is almost never the actual story. Sitting tight remains the highest-probability move — but know that a bad quarter can send this 10-15% lower fast, and you should be comfortable with that. If you're not, sizing matters more than timing.

• If you're trying to buy the pre-earnings dip: The risk here is being early on the wrong side of a "beat and sell" reaction. If you like the long-term thesis, dollar-cost averaging into a position over time is more honest than trying to time the exact bottom — that strategy tends to work better than most people admit.

• If you're watching Broadcom (AVGO) as a related trade: The Marvell-Alphabet news matters, but a strong Nvidia earnings report would likely lift the whole chip sector and give Broadcom some relief. These stocks are correlated enough that Wednesday's print has implications beyond just NVDA.

• For anyone with S&P 500 index exposure: Nvidia is one of the largest weightings in the index at this point. A 10% move in either direction has real portfolio impact even if you don't own the stock directly. Wednesday night matters for the broader market.

• A threshold to watch: If NVDA closes above its 50-day moving average in the session following earnings, that's historically been a sign the selling was short-term positioning noise rather than the start of a structural downtrend. Below it, and you might want to be more patient before adding.