Have you ever watched a stock that feels absolutely unstoppable and suddenly wondered – what if the party is actually nearing last call?
That’s exactly where my head went last night when I saw a note from the only major analyst left with a straight-up sell rating on Nvidia. While literally dozens of his peers keep pounding the table with “strong buy” recommendations, this guy is sticking to his guns and saying the stock could drop another 20% from here. Bold? Crazy? Or maybe the one person brave enough to say the emperor’s new clothes are looking a bit thin?
In a market where Nvidia has basically become synonymous with artificial intelligence, calling for a sizable pullback feels almost heretical. Yet the arguments he laid out were surprisingly detailed – and honestly a little unsettling once you dig in.
The Lonely Bear in a Sea of Bulls
Out of more than sixty analysts covering the company, fifty-nine are telling clients to buy or strongly buy. Six are neutral. And exactly one – from a boutique firm that specializes in semiconductors – is waving the red flag with an outright sell.
His price target sits around $140, roughly 21% below where shares closed last Friday. That would wipe out a big chunk of this year’s gains and bring the stock back to levels we haven’t seen since early summer. In other words, he’s not just trimming his sails; he’s calling for a legitimate retreat.
So what makes this analyst so pessimistic when almost everyone else is still riding the AI wave?
Competition Is No Longer Theoretical
For years the narrative was simple: Nvidia has a near-monopoly on the chips that matter for cutting-edge AI training. Customers had almost no choice but to pay whatever Nvidia asked, and the order backlog stretched out for months.
That story is starting to crack. Hyperscalers – the giant cloud providers who buy the overwhelming majority of these chips – are no longer content to be price-takers.
- Google has been quietly perfecting its in-house TPU chips for years and now openly markets them to third-party customers.
- Amazon continues to expand its Graviton and Trainium offerings.
- Microsoft is rumored to be working closely with several partners on alternatives.
- Even startups like Groq and Cerebras are picking off niche workloads.
The analyst points out something I hadn’t fully appreciated: Google’s latest TPU generation is already beating Nvidia hardware on several important metrics for inference – the part of AI that actually makes money today. And because Google controls the software stack, it can optimize the whole system end-to-end in ways Nvidia simply can’t match inside someone else’s cloud.
When your biggest customer starts telling the world they have a cheaper, faster alternative, that’s worth paying attention to.
Creative Sales Tactics or Red Flags?
Here’s where things get spicy.
The analyst highlights a series of what he calls “sales mechanisms” that Nvidia has leaned into heavily over the past year. These aren’t illegal or even particularly shady on their own, but taken together they paint a picture of a company working overtime to keep growth humming as traditional demand drivers mature.
“We do not see anything malicious taking place… we think Nvidia faces growing competitive pressure and has been leaning hard on a variety of sales mechanisms to adapt.”
One of the biggest eyebrow-raisers is the $26 billion in prepaid cloud compute capacity the company has booked. Management says this is largely for internal R&D and for building out their own DGX Cloud offering. Fair enough on the surface.
But the analyst isn’t buying that the entire amount is truly for internal use. Instead he suspects a sizable portion functions as de facto rebates or “backstops” for major customers. Translation: if a cloud provider over-builds capacity betting on Nvidia demand that doesn’t fully materialize, Nvidia has quietly agreed to eat some of that excess capacity by buying cloud time itself.
It’s a clever way to keep partners happy and orders flowing, but it’s also a hidden cost that doesn’t hit the income statement the same way. And once you start down that road, it’s hard to stop.
The Investment Spree That Keeps Growing
Another trend that caught my eye: Nvidia has turned into something of a venture capital powerhouse almost overnight.
They’ve already deployed $6 billion this year into private companies – many of them potential customers – and have commitments for another $17 billion more. That includes a highly publicized $5 billion deal with a major foundry partner and whispers of an even larger arrangement with one of the leading AI labs that could eventually top $100 billion.
The official line is that these are strategic investments that will pay off handsomely as portfolio companies grow and buy even more Nvidia hardware. And that might very well be true in the long run.
But in the short to medium term, it’s real cash leaving the balance sheet today for hardware sales that may or may not materialize tomorrow. When competition is intensifying, the price of customer loyalty goes up. Simple as that.
Working Capital Tells a Different Story
Most investors look at Nvidia’s massive working capital build and see rocket fuel – proof that demand is so strong customers are literally paying early.
The bear sees something else: evidence that the company is extending unusually generous terms to keep the growth narrative alive. Longer payment terms, bigger upfront commitments from Nvidia itself, inventory build-outs at partners – all classic techniques when pure organic demand starts to soften.
I’ve watched this movie before in other tech cycles. It rarely ends with the stock sailing smoothly to new highs.
Valuation Reality Check
Let’s be honest – even after the recent dip, Nvidia still trades at nosebleed multiples. Forward price-to-earnings north of 40x, price-to-sales around 35x. Those are the kinds of numbers you only get away with if growth stays pristine forever.
Any hint that growth is merely mortal again, and the multiple can contract fast. A 20% haircut in the stock price could happen from valuation compression alone even if earnings keep chugging along at a still-impressive clip.
Add in the possibility that earnings themselves face headwinds from increased competition and higher customer subsidies, and suddenly that $140 target doesn’t look quite so outlandish.
What History Teaches Us
Markets love a monopoly until they don’t.
Cisco in the late 1990s. Intel in the early 2000s. Qualcomm in the smartphone era. All dominant, all seemingly unassailable, all eventually faced meaningful competition that the market refused to price in until it was too late.
I’m not saying Nvidia is destined to become the next cautionary tale. The AI build-out is real and will likely run for years. But the days of 80% gross margins and zero viable alternatives might be quietly slipping away.
The Bottom Line
Look, I own Nvidia shares myself. I’ve enjoyed the ride just like everyone else. But when literally one analyst out of sixty-plus is willing to stick his neck out with a sell rating and backs it up with this level of homework, I pay attention.
Maybe he’s early. Maybe he’s wrong. But the trends he’s highlighting – intensifying competition, creative revenue recognition, ballooning strategic investments – aren’t imaginary. They’re happening right now, in plain sight if you know where to look.
At current prices the market is still pricing in perfection for years to come. Perfection is a tough bar for any company, even one as phenomenal as Nvidia has been.
Perhaps the most interesting question isn’t whether this lone bear is right or wrong today. It’s whether the other fifty-nine bulls are still asking the hard questions – or if they’ve simply stopped looking for storm clouds because the sunshine has felt so good for so long.
Either way, it’s probably a healthy reminder that even the mightiest market darlings eventually face gravity. The only unknown is when.
