This is one of several ads we’ve seen over the past few months that basically say, “AI is awesome, but NVIDIA is going to collapse and some other secret stocks will be the next big winners.” Here’s how the pitch of this presentation begins:

“FORMER MICROSOFT EXEC: ‘Nvidia Could CRASH on February 28’

“The AI chipmaker is facing a potential disaster very few investors are aware of. It could kill the progress of artificial intelligence, crash the S&P 500, and even drive our country into recession.”

The “presentation” is hosted by Chris Graebe, who says he’s “an investor who specializes in promising AI startups.” He runs a “private investing” service at Weiss, one of several newsletters that recommends crowdfunded investments and similar low-level “VC” investing ideas, and I think he used to be a reality TV contestant… but this ad is for the more mainstream Disruptors & Dominators newsletter, which he helps Jon Markman run (being sold at $49/year currently, this is their “entry level” newsletter at Weiss Research), so our picks are apparently from Markman. And they’re probably mainstream tech companies.

Here’s how Graebe refers to Markman…

“I have some extremely disturbing news about the future of artificial intelligence.

“And I got it from a trusted source.

“A former Microsoft executive who worked directly under Bill Gates. And who helped develop early versions of many of the company’s current AI-powered technologies.

“Including a completely autonomous stock picking program, called StockScouter.

“This insider has a massive warning for anyone who’s invested in AI stocks.

“Including anyone who put their money into the AI darling Nvidia.

“My source told me on good information that Nvidia’s stock could crash on Feb. 28.

“Perhaps sooner.

“But the repercussions could go far beyond just Nvidia.”

Markman is a financial journalist, to be clear, when he worked at Microsoft it was mostly because he was a columnist for MSN Money, but yes, he was also the managing editor of that service for a time, having been hired away from the LA Times, so perhaps it’s fair to call him a “former Microsoft executive.” I don’t know how tight he was with Bill Gates. He’s been helming a variety of investment newsletters for years, and took over Disruptors & Dominators fairly recently (it was started by Frank Curzio after he left Stansberry, then was taken over by Tony Sagami for a while after Curzio’s very brief run at Weiss).

It’s been a few years since we covered a Markman tease, I think the last was his long-running pitch for Inseego (INSG) in 2019, which for a brief while was a popular “5G” story stock that might have worked out OK if you managed to ride the surge of demand for home WiFi equipment during the COVID shutdown and sell somewhere near the top, the company was briefly bailed out by that spike in demand for their hardware, but it has otherwise been a disastrous investment if you were convinced of the long-term appeal (INSG is now a penny stock mired in a 1-for-10 “reverse split” and fighting delisting, and perhaps bankruptcy). But let’s see what he’s pitching this time…

The big-picture pitch is that NVIDIA is going to crash because they can’t meet the overwhelming demand for their AI chips, which so many folks persist in calling the “master key” for AI. Here’s more from the ad:

“Nvidia’s AI Master Key is now the most precious resource …

“The hottest commodity inside the tech industry.

“And, well, Nvidia has a big problem.

“In short, they cannot fulfill the swelling demand that has occurred.

“They can’t produce enough AI Master Keys.

Are you getting our free Daily Update
"reveal" emails? If not,
just click here...

“And they may not be able to until 2025 at the earliest….

“If Nvidia is incapable of making exponentially more AI Master Keys immediately …

“The entire AI market …

“The stock market …

“Could all go down the drain.

“And it could happen as soon as Feb. 28.

“If not sooner.”

There’s lots of commentary about how nobody, from Elon Musk to Sam Altman, can get their hands on enough NVIDIA chips, and that has indeed been a challenge for many of the AI startups over the past year, probably stunting the growth of some of them (it’s certainly not money that’s stunting AI growth at the moment — VC investors are throwing as much money at AI startups as they can, even the ones who haven’t done anything yet).

So that’s the challenge Graebe refers to, the fact that if there isn’t enough hardware, the AI projects will wither on the vine. That seems a little extreme, but we’ll see. Here’s how they start to talk up the solution to this problem…

“If more AI Master Keys don’t start hitting the market quickly … many of these well-run AI startups could fail, and fail soon….

“However, there is a glimmer of hope on the horizon.

“You see, my source says there is an emergency plan in place.

“A small group of tech companies — many of them so behind-the-scenes that most Americans have never heard of them — have partnered up with Nvidia to save the day.

“My source, a Microsoft veteran who worked under Bill Gates, has a list of names currently working diligently in the background.

“He calls them Nvidia’s ‘Silent Partners.’

“And they could be the artificial intelligence market’s only hope at bridging the drastic supply shortage of AI Master Keys.

“These companies have banded together to save Nvidia … and the future of artificial intelligence right along with it.”

So what all that leads up to is a tease about Jon Markman’s favorite picks in AI, three companies that he calls “NVIDIA’s Silent Partners.” And that’s what we’ll try to ID for you today… here’s the pitch about the first one…

“The Most Important Company You’ve Never Heard Of

“Nvidia’s First ‘Silent Partner’ has already invested upward of $40 billion in the U.S. alone to help make their heightened production goal a reality.

“That includes a gleaming new 1,000-acre facility dedicated to advanced chips.

“They play such a pivotal role in the supply lines of countless companies that they are sometimes referred to as the ‘most important company you’ve never heard of.’

“They’ve been a critical partner of Nvidia for years.

“Nvidia depends on them to anchor their AI Master Key production.

“And now, with domestic investments surpassing $40 billion …

“They’ve positioned themselves to respond to Nvidia’s distress call.”

What else are we told about this “Silent Partner?” Just this…

“As the supply bottleneck of AI Master Keys starts to subside in the coming months, I expect this company’s role to go public.

“And when they do, they could go up even higher than Nvidia — as the AI market looks for its next champion.

“My source tells me this company is potentially the best chip investment you could make for the next decade.”

Hoodat? You’ll probably be unsurprised to hear that the company Markman is teasing as NVIDIA’s most important “silent partner” is Taiwan Semiconductor (TSM), the largest manufacturer of semiconductors in the world. And yes, they make essentially all of the chips that are designed by NVIDIA (NVDA), which, like many of the sexiest names in microchips, is a “fabless” semiconductor company — meaning they design and sell chips, but they don’t actually fabricate them. That’s outsourced to a fab, and for most people who need high-end chips, that means it’s built by Taiwan Semiconductor. There are competitors, and there are other fabs out there… Samsung is the next biggest competitor, and there are other publicly traded “pure play” fabs, including GlobalFoundries (GFS)… heck, even Intel (INTC) is hoping to catch up and build up its own business of producing chips for other companies… but Taiwan Semiconductor is certainly the world leader, with the most capacity and the most advanced capabilities for building the highest-spec chips.

I wouldn’t say that TSM is particularly “silent,” it’s well-known that they make all the chips for NVIDIA (and Advanced Micro Devices (AMD), and Apple (AAPL), and many others). That led to several folks to pitch TSM as a “safer than NVIDIA” play on the surge of demand for AI chips, since they make most of them… and will make many more, as the chips being designed by Tesla and the next wave of chips being developed by Alphabet and Amazon and others will probably also be built by Taiwan Semiconductor. (Among the ads? Teeka Tiwari pitched them with a “forget NVIIDA, buy Elon’s supplier” spiel… and James Altucher has been hyping them as the “AI Crown Jewel”).

And it’s a very large company — NVIDIA has become by far the largest chip company in the world, thanks to the huge surge in their stock prices this year that turned it into a trillion-dollar company, but TSM is still (just barely) holding off Braodcom (AVGO) to maintain it’s second place position, with a market cap of almost $600 billion. And Taiwan Semiconductor had good news for investors last week, as they forecast that they’d rebound from their disappointing 2023 to post 20% revenue growth in 2024, largely because of the demand for AI chips (AI chips were certainly in high demand in 2023, too, but it took time for TSM to retool capacity to build more of those chips… and at the same time, the other chips they were making, like the processors for Macs and iPhones, were facing much lower demand).

That bit of optimism from TSM management was a market driver last week, so I mentioned it in the Friday File for the Irregulars to close out the week… here’s what I said, in case you missed that piece:

The piece of news that caught my eye this week, though it doesn’t particularly impact my portfolio right now, is the sudden return of optimism to the management team at Taiwan Semiconductor (TSM) — they’ve been suffering over the past couple years as the market for chips stagnated in most areas, with a bit of relief coming as demand or NVIDIA’s AI chips surged last year (though that remained a fairly small part of TSM’s revenue)… but now they’re forecasting 20% revenue growth in 2024, putting them back in the “growth” game after 2023’s significant decline in earnings (we don’t have the final numbers for 2023 yet, but the first three quarters saw revenue shrink by 4.66%, 13.79% and 14.5%, so it’s almost certain that 2023 will be their first full year of falling revenue since 2009… though 2019 came close, with only 1% revenue growth for the year).

Semiconductors are relentlessly cyclical, of course, that’s the very nature of the chip business… and this doesn’t mean that TSM is back to the glory days just yet, growth was well above 20% from 2020-2022 as demand for computers and phones and gadgets soared during the COVID lockdowns, and they’re still facing challenges as they open up new foundries, particularly in the US, but it’s a good sign for the whole industry, and the shares immediately jumped to reflect that. TSM remains somewhat reasonably valued at about 20X earnings (valuation has been as low as ~10-12X earnings during past troughs, including a year ago, but we’re certainly a long way from the cyclical highs — when earnings were booming in 2020 and 2021, TSM got up to ~30X earnings). So those who want that somewhat less volatile exposure to the massive demand for AI chips from NVIDIA, AMD and others, well, TSM is still a rational option. Though I’ll just remind you not to overestimate the safety of even the relatively reasonably valued TSM — both of these stocks are subject to the same big swings in sentiment. Here’s how TSM and NVDA shares looked if you bought them at their early-2022 peaks, both had almost identical collapses at the lows, down roughly 60%:

Past may well not be prologue, but it’s worth paying attention — if a stock is very popular and subject to big sentiment shifts, it’s better to at least be able to get some strong growth out of it when times are good. That’s what makes the big downturns survivable.

So now the forecasts for 2024 earnings have bumped up a bit more, giving investors hope that TSM will be back to close to the $6.50 in earnings per share they posted back in 2023, and will be able to keep growing at something close to that 20% clip in future years. If they can do that, then the valuation is probably still pretty rational — TSM shares are changing hands now at about 18X forward earnings, which is a very reasonable price to pay for a company that is likely to grow earnings at 15-20% per year.

Of course, TSM has been pretty rationally valued — or even cheap — for most of the past 15 years. Investors have been reluctant to assume that future growth will be steady, or bet that TSM will be able to earn more of the returns from the hot chips they produce, or, more recently, have been worried by the geopolitics of Taiwan, where most of TSM’s facilities are located (they are building new fabs in the US, too, though these big semiconductor foundries take several years to build and even longer to staff up and bring online, and they own a couple foundries elsewhere in the world, including mainland China, but the vast majority of their facilities are in Taiwan, and that makes everyone worry about a Chinese blockade or invasion… or heck, even earthquakes or volcanoes, which aren’t foreign to the region).

We’ll see how it goes. TSM’s operating margins have been at decade-long highs recently, as NVIDIA and others have invested in expanding their capacity, and if that continues, and they squeeze more of the cash out of the latest generation of hot AI chips, then perhaps they’ll perform even a bit better than expected. It remains a rational idea, to my mind, and it’s also still a stock I don’t own.

“What are the other ‘silent partners’ teased by Markman? Here are our clues…

“Nvidia’s Second ‘Silent Partner” will also aid Nvidia in the increased production of AI Master Keys …

“This company calls itself a total IT solutions provider.

“They make products that feed other AI developers like cloud solutions, data storage and servers.

“That means they help make the mass rollout of generative AI systems like Nvidia possible.

“They’re an indispensable link in the AI chip production chain.”

That’s not a super-detailed tease, but the best solution from the Thinkolator is Super Micro Computer (SMCI), which is indeed a “total IT solutions provider” that essentially helps to turn NVIDIA’s chips into working equipment, through the sale of servers and racks and other tools that make data centers work.

And SMCI has been one of the true breakout stars of this AI surge — the stock has essentially followed the same path as NVIDIA (NVDA) since the launch of ChatGPT late in 2022… but it is a dramatically smaller company, and therefore the stock’s movement has been far more levered than even the wildly surging NVDA. Here’s what that looks like on the chart, this is SMCI in purple and NVDA in orange, going back to November of 2022… I added Taiwan Semiconductor (TSM) in green and the S&P 500 in blue, just for some context:

Since in many ways SMCI is essentially piggybacking on demand for NVDA’s chips for AI, they’re selling the servers and systems that incorporate and support those NVIDIA chips in data centers, we’d assume that they’ll probably continue to be levered to NVDA’s results — if NVDA keeps having explosive demand, probably SMCI will, too, and it will hit them harder because they’re starting smaller. If we end up reaching a point where demand for AI hardware collapses, or drops in price because the market becomes saturated after the big investments in capacity over the past year, then it will probably drop harder and faster for SMCI, too, that’s usually how it works. If you’re going to ride the rollercoaster, you have to be ready for the scary drops as well as the fun stuff.

I do not have much of a handle on how competitive SMCI’s segment of the market is, or how fast that market can react to demand surges like we’ve seen this year, and that’s probably the biggest risk area… but right now, at least, the analysts that follow SMCI are not at all concerned — they believe that earnings will go from about $12 per share last fiscal year (they’ve just finished their second quarter of FY24) to almost $18 this year, and $22 in the next year… so that’s obviously very good growth, even though those same analysts do think that margins will come down a little bit in the future (from 11-12% EBITDA margins to ~9%). So that’s future growth in the 20-25% range, and the stock, at $445, is valued at about 37X trailing earnings (adjusted), and about 21X what they’re expected to earn over the next four quarters. Pretty reasonable if that growth is as good as expected and the market stays strong… and SMCI was doing pretty well before the AI mania heated up, they roughly doubled their revenue in the two years before we all went AI-crazy in early 2023, so perhaps that growth will be in some way sustainable.

I’ve written briefly about SMCI in the past — most recently as a play on the growth in liquid cooling for data centers — but I’ve never owned the stock or become comfortable with guessing whether they can hold on to their market share in the “servers and racks” business, which I assume is likely to be very competitive. If I’m too cautious on that front, and SMCI can hold on to their market share during this ongoing surge of AI investment, it may end up being a great investment — the company’s still pretty small, with a market cap of about $24 billion, and they’ve certainly had a few years of very strong revenue growth.

Though I can’t see how SMCI will be able to help cushion the blow if NVIDIA (and TSM) can’t make enough of the high-end AI chips that the market is craving right now. I’d say SMCI is riding the AI hardware wave along with NVDA, not really “helping” or “partnering with” them.

And there’s one more “silent partner,” and the hype is strong with this one…

“There’s one that stands out from all the others.

“But they don’t lift a finger in the actual physical production of Nvidia’s most advanced chips.

“You see, every AI Master Key needs a brain.

“A series of instructions and designs that need to be executed perfectly.

“And this company already supplies the brains for the technology used by billions of people.

“In fact, at least 70% of the entire world’s population already relies on them every single day, without even knowing it.

“Everything from supercomputers to smartphones to automobile safety systems requires their foundational technology.

“I’m talking about Amazon, Apple, Google, Intel … a who’s who of the world’s leading tech corporations, they depend on the essential blueprints they provide.

“This company has their hands in virtually all current tech megatrends.

“Including edge computing, robotics and the internet of things.

“Bringing in more than $700 million in quarterly revenue.

“Now, they are underpinning AI, the greatest tech megatrend of the 21st century …

“Enabling it to work everywhere.

“Their tech is at the heart of nearly all AI devices available today.

“And for Nvidia’s AI Master Key to function properly moving forward …

“It’s imperative that this company’s designs play a vital role.”

That’s actually a tease for a company that NVIDIA tried to acquire a few years ago (they were rebuffed by regulators) — here Markman is hinting at ARM Holdings (ARM), which is the developer of the core ARM architecture used by most chipmakers. They both sell chip designs and license their basic architecture to companies who want to design their own chips, so in many ways this is a critical and foundational company in the semiconductor business, and it’s also really just an intellectual property and design house, so their revenue almost all comes in the form of licensing fees and royalties, which means they have exceptional margins.

After the failed NVIDIA takeover, lead investor Softbank took ARM public at $51 per share back in September, and it has done well so far — it was a very popular IPO, so the first few trades were at a rich premium in the $60s, then it fell back to $50 or so in November, and lately, with the market’s recovery and the enthusiasm for AI and technology in general, it’s been ticking close to $80 per share. That means the stock trades at a very optimistic valuation — the forecast is that they’ll earn about $1.06 in their current fiscal year (adjusted earnings, excluding stock-based compensation) and will grow earnings at a 20-25% pace for the next couple years, so that means the stock is valued at about 70X current adjusted earnings. It’s an exceptional company… but that’s a pretty exceptional valuation, too.

I can’t justify buying ARM at the moment, but if I were so inclined I’d probably want to wait until after the lockup period expires and the insiders are allowed to start selling their shares, that’s often a point when recent IPOs experience some selling pressure. For ARM, the six-month lockup period should end on March 12. No guarantees that the stock will fall at that point, of course, maybe the story will heat up and the shares will surge even higher, but that’s the time period when I’d start looking if I wanted to buy ARM. Which at anywhere this valuation, I don’t.

Lots of folks do, though, including Taiwan Semiconductor, which invested in ARM at the IPO, so I could be too cautious. We’ll see how it turns out. The IPO was pretty limited, Softbank sold a pretty small stake at first, as is somewhat common (best to try to generate interest when the supply of shares is limited), and still owns 90% of ARM… so it’s also possible that if pricing stays elevated, we’ll see Softbank trying to “monetize” a much larger portion of their ARM shares.

And there you have it, dear friends… some worries about whether NVIDIA and the AI market might “collapse” on February 28, which is a rough approximation of when NVDA will release it’s next quarterly earnings report and likely update investors on their forecast for AI chip sales in 2024 (or at least in the next quarter)… and it’s true, if NVDA’s results or outlook end up being disappointing in a big way, then NVDA shares and many other tech stocks are likely to fall pretty hard.

NVDA brought the AI party to the market, and they could be the first one to take it away… though I don’t expect it will be because of their inability to make enough H100 or M100 chips with Taiwan Semiconductor and other manufacturing partners, if NVDA falls hard it’s much more likely to be because the demand is eventually sated, and revenue growth slows down and brings with it lower prices and lower profit margins. That seems unlikely to hit at their next earnings report, but it’s a long way from being impossible… and it’s especially impossible to know when investors will begin to feel more cautious about NVDA shares. So far, all we can say is “not yet”, as NVDA has spent pretty much every trading day so far in 2024 hitting new all-time highs.

Will NVDA’s “silent partners” be the ones to save the AI rally? Or will they collapse like NVIDIA at the first sign of saturation for the AI chip market? Again, I dunno… but at this point, at least, Taiwan Semiconductor and Super Micro Computer are priced pretty rationally given their recent growth and their expected growth rates, while ARM Holdings is in a different category and is probably still a bit more speculative, given its very rich valuation as a recent IPO, despite having a much more profitable business model and a near-monopoly in some areas of semiconductor architecture.

I’d guess they’ll all “ride or die” together as the AI mania continues, perhaps leading to the formation of a real bubble this year… but the mania could always deflate before a real bubble inflates, and I imagine everyone will be watching the earnings from Microsoft (MSFT) and Meta (META) and Advanced Micro Devices (AMD) next week for clues about how the AI chip market is looking, since META and MSFT are both major NVIDIA customers and are spending so heavily to build out cloud AI capacity, and AMD is NVDA’s most likely near-term competitor in the high-end AI chip market. And I’m sure we’ll all be watching extra-closely when NVIDIA itself reports a few weeks later (AMD and MSFT will report on Tuesday next week, META on Thursday… NVDA has not yet confirmed their release date, but it will probably be around February 22, which is when they reported last year).

And that’s all I’ve got for you on Markman’s tease about these “silent partners” for you today… have a favorite in the bunch? Think others will be better bets for whatever the next era of AI might bring in 2024? Let us know with a comment below.

Disclosure: Of the companies mentioned above, I own shares of NVIDIA. I will not trade in any covered stock for at least three days after publication, per Stock Gumshoe’s trading rules.