Shah Gilani has been promoting “next NVIDIA” AI stories over the past year, both for his previous publisher and for Manward Press, where he recently took over as “lead pundit” from Andrew Snyder… but this one is pretty juicy, so lots of readers have been asking about it. This is all from a sales pitch for his entry-level newsletter, Manward Money Report ($99/yr.).

Before we get into the full story, I experimented with an AI-generated video to summarize some key points from this teaser reveal, so you can check that out first if you like… it’s accurate, but of course has a lot less detail than the article… let me know if it’s helpful or just annoying.

Here’s how the ad starts:

“This little-known AI startup could dominate the potential $25.6 trillion per year AI market…

“Discover how you can get in today before its potential 4,735% revenue surge in 12 months….

“While Nvidia is making headlines for its record earnings…

“And more and more people pile into the stock…

“Most people are missing out on an even bigger story.

“I’ve uncovered a Nvidia partnership that’s been quietly tucked away.

“The general public knows little about it.

“You won’t find much information on it in mainstream sources…

“But today, we’re breaking this overlooked story… and the company behind it, which I call ‘Nvidia’s Secret Partner.'”

So… is this one of the 50-plus different AI stocks we’ve covered over the past year, or something new? Certainly plenty of companies have been pitched with that “NVIDIA partner” angle, or as secret suppliers to NVIDIA… or even as the companies that might “kill NVIDIA”, and that chatter will no doubt pick up as we close in on the NVIDIA earnings report (which is in about two weeks)… but let’s check the clues in the ad and see what the Thinkolator can tell us…

This is apparently a company that is directly involved in the use of NVIDIA’s chips in data centers…

“This company is just a fraction of Nvidia’s size … but as you’ll see, it is crucial to the success of Nvidia going forward.

“Because, as I’ll show you in a moment, this company’s new technology solves THE biggest problem in the AI chip industry….

“Every big player in the industry uses it, including… Nvidia… Intel… AMD… Microsoft… And Amazon.”

And it sounds like this is one of the “invest and partner” companies Amazon has worked with — they like to make long-term buying deals, but also use their purchasing power to get a little equity in the company they’re partnering with, too. We’ve seen that with their past deals for air cargo (Cargojet) and electric delivery trucks (Rivian), so I guess it shouldn’t be a surprise that they’ve made similar deals with some partners on the Amazon Web Services side, too.

“Amazon just invested $144 million in this startup…

“And agreed to buy at least $650 million worth of its products in the not-to-distant future.”

We’re also told that the company is roughly the size NVIDIA was in 2014 (after which, we might recall, it eventually rose something like 10,000%)… so that would mean it has a market cap in the neighborhood of $10 billion. Not all that small for a “startup,” but certainly a lot smaller than the big AI names being bandied about.

And just in case the “like NVIDIA” FOMO promise wasn’t clear enough, Gilani spells it out:

“In a few weeks, I believe there will be a very important announcement…

“One that could catapult the company into the mainstream and send its shares soaring into the stratosphere.

“That’s why I’m sharing details on this new opportunity with YOU now.

“Because there is a real chance you could ride this NEW AI investment to hundreds of dollars per share in the coming years… just like early investors did with Nvidia.”

We’re also told that he found this “next blockbuster AI stock” because of one of its venture capital funders…

“I first discovered it while reviewing the investment portfolio of a small and unusual venture capital firm.

“A Very Unusual Venture Capital Firm Quietly Put a HUGE Position in This Stock.

“It’s called Sutter Hill….

“… most VCs will invest in a lot of startups… and just need a few huge winners like Google or Facebook to bring in 10x returns or more.

“Sutter Hill, however, does the EXACT OPPOSITE.

“Instead of investing in a hundred good companies…

“It invests 80% of its money in just ONE great company.

“So, for example, Sutter Hill decided to back Nvidia all the way back in 1999 before anyone had ever heard of it.

“But they were so confident in Nvidia’s future, that they decided to basically bet their business on its future.”

Gilani says that Sutter Hill owns 12.6% of this company, which he says is “massive” — so there’s another clue.

And the company is apparently involved in building connections inside data centers in some way… the tease says that NVIDIA’s new Blackwell chips, which were the talk of the town when they were introduced earlier this year, are so powerful that they require newer and better systems to connect them… here’s how he puts it:

“Blackwell is too advanced. And too powerful.

“All the technology connecting these chips together cannot handle the amazing power and speed of Blackwell.

“Without the ability to send these 20 quadrillion calculations at lightning-fast speeds, it’s like driving a Lamborghini on a one-lane road in rush hour.

“That’s where ‘Nvidia’s Secret Partner’ comes in.

“They’ve built new technology that connects all the chips in these AI clusters.

“Their technology doubled the data transfer rates.

“It cut energy usage in half.

“Plus, it can clean and amplify signals, increasing the reach of connections over long distances.”

And some specific hints about the company’s performance so far…

“From 2021 to 2023, revenues grew from $34.8 million to $115.8 million… that’s a 232% rise…..

“… as Nvidia continues to dominate the AI chip market… and the world’s largest companies rush to buy its products…

“‘Nvidia’s Secret Partner’ will ride this wave of cash and could become one of the most explosive stocks in the AI market.

“That’s why I predict this single stock could be THE top-performing AI stock in the coming months.”

And it’s not just NVIDIA that he uses as FOMO fuel, he name drops a few other big winners from past booms as well…

“Cisco made it possible to connect to the internet…

“Akamai made it possible to stream videos instantly on the TV…

“Broadcom made it possible to turn your phone into a computer…

“Now, Nvidia’s Secret Partner is making it possible for anyone to tap into game-changing AI technology.”

And that “big announcement” that Gilani hints at is really just, in my words, “it will become clear that they’re partnered with NVIDIA, and investors will go nuts.”

Here’s how he puts it:

“Right now, the partnership between this company and Nvidia has been kept under wraps.

“Few people outside the industry know about it.

“But I’ve been able to get my hands on their IPO documents…

“And found something very interesting.

“The IPO documents say that ONE large company is the main reason why their revenues have surged 232%…

“But they don’t actually say who it is.

“And while they won’t confirm anything…

“The writing is on the wall.

“Nvidia is listed FIRST on their IPO as their premier partner.

“And I discovered that Nvidia Chief Executive Jensen Huang personally endorsed this company during its roadshow.”

I have no idea whether this startup will talk about NVIDIA on their next earnings call, and let “mainstream America” know that they’re connected, or if the stock will go ballistic as a result — that’s certainly possible, and stocks have soared for much less “real” reasons in the past year (remember when every stock that just had “AI” in its name doubled almost overnight, in the early enthusiasm about ChatGPT?).

But, of course, the institutional investors won’t be surprised… and investors who have been paying attention won’t be surprised, they’ve already bid this stock up to a pretty wild valuation in its first few weeks as a publicly traded company. Shah Gilani is teasing Astera Labs (ALAB), which just went public on March 20.

And as luck would have it, ALAB reported their first quarter earnings yesterday… and they did talk quite a bit NVIDIA on the conference call, and highlight their plans to help support those new Blackwell installations late this year with their new connecting hardware and software, though that is really all “known”, at least in the broad strokes (and ALAB did demonstrate their new products at NVIDIA’s conference in March — and make clear, in their IPO filings, that they collaborate closely with the key infrastructure providers like NVIDIA, Intel and AMD). Still, the stock drifted lower after that news came out, so if it’s some “mainstream America will get excited” surge you’re looking for, well, it at least hasn’t happened yet.

Or happened again, I should say. ALAB did get a crazy pop when it went public, and the IPO was at a crazy valuation to start with, so clearly the lust for a decent-sized company connected to AI hardware in some way drove that very successful IPO — they sold their shares at $36 in the IPO on March 20, and the stock just about doubled to $64 or so that day, then soared to $95 at the highs in the first week, but it’s now back to the low $70s. So the stock today, after falling about 5% after earnings, is roughly in the middle of its trading range during this first six weeks as a public company.

And their earnings report did look quite impressive, as you would probably expect for an IPO in a hot area of the market — their quarterly revenue grew by 269% over the past year, and they’re talking up their latest generation of their connectivity products.

What are those “connectivity products?” I can’t say that I have any real understanding of how this stuff works, or of what the competitive landscape looks like for this stuff, but here’s how CEO Jitendra Mohan described the situation and their products in the quarterly earnings press release:

“Astera Labs started the year strong, achieving record revenue in the first quarter, driven by the accelerating deployment of AI infrastructure. As hyperscalers embark on a significant transformation of their data centers to support AI applications with increased capital investment, we’re witnessing the emergence of a multi-year growth cycle. Our Intelligent Connectivity Platform, comprising of the COSMOS software suite and semiconductor-based PCIe, Ethernet, and CXL solutions, is uniquely positioned to support this growth and is foundational to deploying AI infrastructure at scale. In the first quarter, we further extended our connectivity platform and started sampling our third generation of Aries Retimers with support for PCIe 6.x and the industry’s first PCIe/CXL Smart Cable Modules for Active Electrical Cable applications to enable multi-rack GPU clustering.”

I probably wouldn’t know a “Retimer” if it bit me on the ass, but it at least sounds impressive. And their revenue growth provides at least some level of endorsement that “this stuff is in high demand.” There’s also some meaningful potential that because they’re focused on high-end installations with large clusters of GPUs, and say they’re in the lead when it comes to the next generation of this connectivity stuff, they could have meaningful growth connected to the installations of those new NVIDIA Blackwell GPUs starting late in 2024. So if you’re looking to do the mental gymnastics required to justify the current valuation as we go through these numbers, keep that “growth might get better in a year or so” idea in the back of your mind — you’ll probably need it.

Once you get past that top line and their impressive revenue growth, evaluating the business gets much tougher — that’s because they’re not sustainably profitable yet, and in the first quarter that was skewed dramatically by the cost (in both new shares created and listing costs) of the IPO. The products themselves at least make financial sense, apparently, they’re selling them with a 70-80% gross margin, so that’s good, but they have so much share issuance that their GAAP operating loss ($83 million) in the first quarter was far higher than their revenue ($65 million).

The good new is that their adjusted numbers, if you ignore that massive stock-based compensation and the impact of the IPO, look a lot better — on an adjusted basis, they had 78% gross margins, operating income of $15.9 million, net income of $14.3 million, and “pro forma non-GAAP diluted” earnings per share of ten cents. So they were sort of profitable… and think the growth is going to keep booming, though the impact of a much higher share count will mean that adjusted earnings growth could be pretty minimal (their guidance is for non-GAAP diluted earnings of 11 cents per share — largely because they’ll go from 153 million shares outstanding before the IPO last month to about 180 million by the time they report again).

They did get a nice cash infusion, since most of the shares sold in the IPO were sold by the company (about 15% were sold by their existing shareholders), so ALAB now has about $700 million in cash to fund the next wave of its growth… but there could also be a wave of selling pressure as the insiders get through the initial post-IPO trading embargo (which in their case ends two days after they report their second quarter earnings — that probably means insiders can really begin selling in the second week of August).

And yes, Sutter Hill is a meaningful shareholder of Astera Labs — they participated in their equity raises in 2021 and 2022, generally buying preferred stock that was converted into common stock at the IPO, and also bought some shares from executives over the past couple years when those employees wanted liquidity. The only other big named venture backer of Astera is Fidelity, and Sutter Hill was the biggest shareholder, at 13.7% of the company pre-IPO (those shares also show up under Director Stefan Dyckerhoff, who is Sutter Hill’s representative on the Astera board).

Sutter Hill is not infallible, of course, but they have a fascinating history and are very concentrated compared to a lot of other venture capital companies — they’re one of the original VC firms, and have been around for 60 years now, but today they’re probably best-known for helping to build Snowflake (SNOW) and Pure Storage (PSTG) (good piece here from Jeff Burke on substack about Sutter Hill and what makes them unique, if you’re interested). I would assume they’ll end up gradually selling down the shares after the lockup period, as is typical for venture-backed IPOs.

So… will Astera Labs be a great investment from here?

I don’t know. Fast-growing company, but trading at a very optimistic valuation.

It’s very early on, this is still a very young company (founded in 2017), though it has experienced and connected founders and strong backing from Sutter Hill and others, and in some ways it reminds me of a younger Arista Networks (ANET), in trying to innovate with better software-based connectivity equipment that appeals particularly to the “hyperscale” data center operators — which means they’ll probably continue to have a very concentrated customer base, which increases the quarter-to-quarter risk as those customers place or pause their orders. In 2022, they got 89% of their revenue from their top three customers (and more than 50% from one customer), and in 2023 those top three customers represented about 70% of their revenue. The companies are not named, but it would be surprising if they weren’t some subset of Amazon, Meta, Microsoft and Alphabet (and yes, Amazon does have a partnership deal with Astera, so they have committed to a base level of spending and have warrants to buy about 2.3 million shares for $20.34 each through 2029 — they’ll earn those warrants by spending a total of $650 million on Astera’s products over many years).

It’s an impressive company, but with 180 million shares outstanding by the end of the second quarter (their guidance), we’re also seeing it trade at a pretty impressive valuation — that would mean a market cap of a little over $12 billion, for a company that is marginally profitable on an adjusted basis and probably will continue issuing a lot of shares in these first few years… and any real GAAP profitability is likely to be a ways in the future. It could transform into a sustainably profitable company fairly quickly if things go well, particularly if their hyperscaler customers continue to order heavily, but at this valuation investors will probably be looking for continued nosebleed-level growth.

As of today, Astera Labs has posted sales of $98 million over the past four quarters, and is valued at about 120X those trailing revenues. That’s about as extreme as you get, so to take them seriously you do have to give them some credit for the very rapid revenue growth. If revenue keeps growing at 10-12% per quarter, sequentially (not year over year, but one quarter to the next), as they are currently guiding for the second quarter, then their revenue could top $300 million this year, or perhaps even a little above that. We’re obviously guessing once we get past the horizon and talk about the future, but the end market is certainly growing strong as investment in data centers remains aggressive for all of the hyperscalers, and their products have so far been well-received. If they get above $300 million in sales for 2024, then they’re valued at something like 35-40X forward revenue.

That puts them in extremely rarefied air, so they’ll have to be pretty special for this to work, and keep growing at a similar rate for more than just this year — once you get outside of biotech and a few uranium miners, almost nobody trades for more than 30X forward revenues — that’s where you find some serious outliers like Microstrategy (MSTR), which really is traded as a bitcoin proxy, nobody cares what the revenue or earnings of the actual business are, or the aspirational flying electric taxi company Joby Aviation (JOBY). They do have some company among the more established companies, if you’re sorting for firms that are valued above 35X trailing revenues — ARM holdings (ARM) has ludicrously high margins as primarily an IP and licensing company, and they trade at almost 40X trailing sales, and NVIDIA itself is at about 37X trailing sales, but it’s a short list of companies which are the focus of extreme investor optimism (and ALAB will only be at 35X trailing revenues if they get well above that $300 million revenue number… and continue to have a share price near the $60-70 range a year from now).

And just as I was about to post this article, the analyst estimates started flowing in — so we can tell you what they think, too (analysts wait roughly a month before initiating coverage of a newly public company). They’re slightly more pessimistic than I, they see an average of $277 million in 2024 revenue for ALAB, growing at 45-50%/year after that, so at $69 that means roughly 43X current year sales estimates. And those same analysts are forecasting 31 cents in adjusted earnings this year, growing at ~70%/year to 90 cents in 2026, so on that front we’re somewhere north of 200X forward earnings. The analysts have price targets generally in the $85-100 range, for whatever that’s worth (not much, to be honest), but the average rating is “buy”, and there’s usually a certain amount of investor money that follows these sell-side analysts, so even though those forecasts are not all that extravagant, the analyst attention could bring a little buying pressure in the next few days.

The rich valuation doesn’t mean the optimism is unwarranted, or that ALAB is doomed… it’s too early to guess at that, at least for me. But what’s pretty clear is that at this price, they have no room for disappointment as they grow. I actually really like the story here, and I’m impressed by the founders, and by how quickly they’ve built this product line and customer base, so I do recommend reading the “Letter from our Founders” part of the prospectus if you want to talk yourself into buying a few shares in these early days, but I’ll hold off for a while. I do occasionally dabble in very new companies, but I tend to find recent IPO’s more attractive once they’ve got six months in, have reported a couple quarters as a public company, and have made it past the expiration of that first insider lockup period.

As always, that’s what I think… but it’s your money, and you get to make the call.

To overthink this a bit more… the comparison company that sticks in my mind remains Arista Networks (ANET), and that might partially be because I’ve written about that one a few times this year, and it just reported a great quarter yesterday. Arista has different products, but essentially the same strategy and business drivers (both build software driven connectivity solutions for the the complex and rapidly-changing demands of large customers who want to speed up communication within data centers), and is also heavily reliant on a few hyperscale customers (though not to quite the degree of ALAB today). So that’s an area of both promise and potential peril — ANET has been public for about a decade now, and has been a great investment, but took some time to get going… and they were also more mature at IPO than ALAB to some degree. ANET had grown sales by about 60-70% per year for the two years going into its 2014 IPO, but was also (unlike ALAB) easily covering its operating and R&D costs and making a profit back then, and it went public at about 10X sales. In its first few years as a public company, Arista Networks was growing its revenue by about 40% per year, was valued at about 50X earnings, and mostly traded at about 10-12X sales. When they hit a $12 billion market cap in 2017, which I focused on because that’s about where ALAB is today, ANET traded at about 10X sales, 50X earnings, and was growing revenue at 45%. It could have worked out for ANET shareholders to buy at 30-40X forward revenue in 2017, but it would have required a lot of patience — that would have valued the company at about $70 billion, and the rest of the market didn’t think Arista was worth that much until about six years later (last Fall, to be precise — it’s now at about $90 billion, after the recent surge).

The ray of hope in that comparison is that ALAB might grow its revenues by 150-200% this year, and if revenue growth stays at 100% plus for a couple years, well above what analysts currently expect, we could find that investors are willing to justify almost any price… and where there’s growth, there’s always hope.

The more apt comparison then, given the valuation, might be Snowflake (SNOW), which is a very different kind of technology company but was also launched by Sutter Hill and went public at more than 100X sales — that was a widely admired company, with a great product and a revered CEO, and a lot of people thought the 100%+ revenue growth could justify the rich valuation… but things only went pretty well, not quite perfectly, and the growth gradually slowed down from 170% at the IPO to about 100% two years later, and then really faltered, so as of today that growth rate has come down to 36%, and the valuation has collapsed. Which means the investors who paid more than 100X sales after the IPO, in late 2020 or pretty much anytime in 2021, have seen their investment fall in value by at least 35-40%.

If Shah Gilani is right when he says, “I expect revenues could surge as much as 4,735% over the next few months,” then sure, the stock will probably go up… but that’s just a stupid marketing tease, not the kind of thing you want to take too seriously. If Astera were a $3 billion company today, I’d be very interested… at $10-12 billion, I’ll resist the urge to speculate. At least for now. I will keep an eye on it, though, and I’m encouraged that it’s at least starting to drop a little bit today.

Comments? Questions? Throw ’em into the friendly little comment box below — thanks for reading!

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