Here’s the latest teaser promise for us to decipher:

“AI Energy — The $40 Trillion Disruptor Bringing Star Power to Earth!… the company with AI boosters, the technology that’s critical to unlocking what’s being called ‘the Holy Grail of energy’ … a chance for investors to literally get in on the ground floor of a brand-new form of energy.”

So what’s the “AI Energy” stock that Ian King things we should buy? It’s all wrapped up in a teaser ad that’s designed to get you to subscribe to his Strategic Fortunes ($79/yr) newsletter at Banyan Hill, which is their “feeder” letter (meaning it’s the one that comes at an “entry level” cost, and is used to build a pool of subscribers who can then be sold “upgrades” to King’s pricier True Momentum or Extreme Fortunes letters).

But let’s short-circuit that sales process and see if we can answer the tease for you without entering that “upgrade” cycle, shall we? He drops a few hints, so the Thinkolator ought to be able to help us out…

The big focus of the story is nuclear fusion… that’s what they mean when they say this disruption is “bringing star power to earth” … and the argument there is that the breakthrough in nuclear fusion that we saw about a year ago, when Lawrence Livermore National Laboratory reached “ignition” with their fusion experiment, is going to lead to what Ian King refers to as the fifth major energy revolution (following fire, coal, oil, and nuclear fusion).

That fusion accomplishment was a big deal, that was the first time that hydrogen atoms were fused together, using a barrage of lasers, and released more energy than the lasers pushed at them. I’m sure it’s not really a “net energy” gain just yet, because that doesn’t include the energy of running the superconductive magnetic field to surround the plasma created by the fusion reaction, and it’s so far just a brief moment of success, not a sustained energy source… but it’s real progress in fusion, and our knee-jerk reaction to that twenty years ago would have been that it’s a hoax, so I guess there’s finally hope for a real fusion energy source in the future. Particularly because they were finally able to replicate that experiment at Lawrence Livermore in July, after a couple failed attempts, proving that it wasn’t a one-time fluke. So fusion might be really coming, even if a lot of experts think it’s still decades away (not that any of them really know, of course — progress and innovation are iterative, and time consuming, but do not always move up a steady slope).

And regular ol’ nuclear fission has been in the spotlight again this year, too, thanks to the need to decarbonize and the political and environmental cost of oil and gas, so we’ve certainly seen plenty of nuclear teasers (Porter Stansberry touting BWX Technologies (BWXT), Whitney Tilson pitching several nuclear companies, Nomi Prins touting NuScale (SMR), etc.) … but we haven’t looked into a fusion stock yet. That’s mostly because there aren’t any publicly traded “fusion” stocks, not really, these are all pre-commercial projects and are mostly run by national labs or multinational consortia, though the billionaires have also funded a few fusion-focused startups that are trying to push things ahead faster.

But Ian King has a way for us to play this fusion breakthrough… so what is it? Really, it’s the AI part of the equation.

Part of the challenge of harnessing nuclear fusion is controlling the reaction and the lasers and the magnetic field (or whatever else you use to contain the plasma), and that has become more feasible, in part, because of supercomputers and, apparently, through the use of artificial intelligence to rapidly adjust to the fusion reaction. So what King is really pitching is one of the companies involved with that part of the project — not the fusion reaction specifically, but the supercomputing tools that are being used to help create and control the reaction. Here’s how they start to hint at that in the transcript for this ad:

“So, in the background of these fusion and AI breakthroughs, an innovative company was developing a technology that takes the computational power of artificial intelligence to exponentially greater levels.

“And that’s where the firm, with 11,000 patents comes in….

“AI energy is going to change the energy markets forever, and there’s also a huge investment angle to this, as there is with any new technology breakthrough.

“This company is headquartered just down the road from that facility where AI energy was born. And this firm has invented a device that ensures supercomputers … are truly super.

“They call it Extreme Performance Yield Computing.

“That’s a mouthful. I call them AI boosters.

“We understand how rocket boosters work. They create enough thrust to propel a space shuttle through our atmosphere and into space.

“This AI booster microtechnology takes the already immense computing power of existing supercomputers — and it augments their AI capabilities. It boosts their performance exponentially.”

And we get some further comments about this secret investment…

“The way I see it… in order for these fusion projects to work — these boosters are needed. Without it, nobody will be able to sustain an AI energy reaction long enough for commercial use and mass adoption….

You can buy into the company that owns that patent pipeline and their tech has become mission critical, virtually irreplaceable for the biggest AI energy projects.

“If someone has $5 to spare — that’s all it takes to grab a stake. If someone has $1,000 or more — they can go in for an even bigger stake.”

And they imply that Lawrence Livermore is indeed using this specific “AI Booster” in their fusion experiments…

“Imagine if you could insert a device into your car and make it 2 million times more powerful — we’d all be driving rocket ships.

“Now imagine boosting the already incomprehensible performance of a supercomputer by 2 million times.

“Earlier you mentioned the facility 50 miles away from Silicon Valley. The Lawrence Livermore National Lab.

“That’s the fusion reactor that was shown on CBS’ 60 Minutes.

“They’re installing AI boosters at this very moment, into the El Capitan supercomputer at that facility.

“This will allow that supercomputer to reach 2 quintillion calculations per second.

“It’s a crazy number. It’s a 2 with 18 zeros behind it.

“What it means is that this supercomputer will become, by far, the fastest supercomputer in the world. Thanks to AI boosters.”

So what’s in those “AI Boosters?” Here we get a little commentary about the company and the product…

“It has a really innovative team of engineers who have worked at firms like Hewlett Packard, IBM and Micron to name a few … and it attained 11,000 patents for this AI booster microtechnology.

“But if you broke one open — you’d find four key components that really do the heavy lifting to make a supercomputer exponentially faster.

“First, is what they call, Infinity Fabric.

“Its role in boosting AI is to keep all the data moving quickly so it doesn’t bottleneck … and keeps the system running smoothly….

“There’s also the processing node … as its name implies, it’s what gives AI the speed it needs to process all those calculations so quickly.

“There are over 39 billion of these nodes in just one of these AI boosters…

“look at the thread relay. This is what makes AI so dynamic.

“It’s what really amps up the speed of AI applications and boosts the speed of the supercomputer running those applications.

“There are 128 threads in 1 of these AI boosters.

“Lastly, we have PCIs … that’s short for Peripheral Component Interconnect.

“Much like the human brain has receptors that quickly transmit signals throughout our bodies … PCIs do the same with supercomputers.

“Rather than send data down a two-lane road … PCI’s open the doors to a massive superhighway that ensures all the data — coming from all the different systems within a supercomputer — never gets stuck in traffic.”

King talks about some other folks who are using this company’s “AI Boosters,” like Mercedes for the design of its Formula One car, scientific teams at the University of Florida, Oregon State, and Notre Dame, and Microsoft (to which King notes, “how long before they use them for ChatGPT?”).

So that’s plenty of clues, no? Yes, the Thinkolator is revved up and waiting… and it’s just a few minutes of cogitationizing before we learn that this is the company that has long been chasing NVIDIA’s shadow in the AI world, Advanced Micro Devices (AMD).

You may well have heard of AMD, a company that has become an incredible success story under CEO Lisa Su — they have been taking share from Intel in CPUs for years now, and have been nipping at NVIDIA’s heels in the gaming market with their Ryzen GPU chipsets, and they’re trying to take some market share from NVIDIA in the booming AI processing business, too, with high hopes for success from their latest data center chipset, the MI300, which is supposed to be shipping in volume next year.

How do we match our clues? Well, AMD’s chips are indeed labeled with that Extreme Performance Yield Computing name, they use the “EPYC” acronym… and AMD is the GPU provider for El Capitan, the supercomputer being built by Hewlett Packard and AMD for Lawrence Livermore (and expected to replace an IBM/NVIDIA machine that represented the supercomputing state of the art five years ago). And they have applied for a little over 11,000 patents, as teased, though I think only about 8,000 have officially been granted at this point.

And yes, AMD’s EPYC chips over the past few years have had more than 39 billion transistors… though I don’t know that most folks would call them “nodes”. AMD’s most advanced MI300 chip and the most advanced (and biggest) NVIDIA GPU chipsets have been in the 150-billion neighborhood by that measure, so I don’t know if that number really means much, but it lets us match the clue given.

The one place where we don’t really match? The share price… King hints that we can “start with as little as $5,” and that’s true, you could invest $5 in AMD today if you’re using a brokerage that offers fractional share trading (which most of them do now), but you’d be buying less than one share. Given the perfect match everywhere else, I’ll assume that’s what he’s talking about.

I would have looked a little more fondly on AMD in the “AMD vs. NVIDIA” debate a year or two ago, because AMD used to have faster growth and a lower valuation than NVIDIA, but these days that’s no longer true. NVIDIA’s big lead in AI chips meant that they absorbed almost all of the surge in demand for data center GPUs this year, and enjoyed a massive revenue and profit windfall as a result, and that means AMD, with the delays in the release of its MI300 that was designed to compete with NVIDIA’s Hopper H100 data center/AI accelerator, has had to mostly just watch and wait, hoping that they’ll get a piece of this boom market next year. They’ve also had to live with the fact that their non-AI CPUs and GPUs are used in a lot of servers and desktop computers that are seeing lower demand this year than in the past few years of COVID-fueled growth (similar to the challenges of other more diversified players in the chip space — Taiwan Semiconductor (TSM), for example, makes all of the high-end AI GPUs that NVIDIA and AMD design and has lately been teased as the “AI Crown Jewel”… but they also make all kinds of semiconductors that go into other products, from automobiles and smartphones to laptop computers and industrial robots, and many of those markets have been in the doldrums, especially the high-volume smartphone market as Apple’s iPhone sales have softened, so TSM’s revenue has been disappointing even though they’re an essential player in this AI boom).

But if this AI hardware demand continues for years, AMD might indeed grab a big chunk of that market… we’ll see — it’s hard to unseat NVIDIA, which has long had a product lead in this area and has focused on almost nothing else for a decade, including building out the valuable CUDA operating system platform which provides somewhat of a competitive moat, but lots of people are trying, and the demand has been so strong that the big customers will have a real incentive to support NVDIA’s competitors, if only to keep the market somewhat balanced and keep pricing under control. (They’re also trying to design their own competing GPU chips for AI processing, so folks like Tesla, Amazon, Alphabet and others are all trying to wrest that market from NVIDIA… but again, it ain’t easy or fast, and most of the companies trying to design their own chips for tomorrow are also trying to buy as many NVIDIA GPUs as they can get their hands on to use today.)

Right now, analysts expect AMD to have about $3.32 in adjusted earnings over the next four quarters, so at $117 they’ve got a forward PE of about 35. By that same metric, NVIDIA, which is growing dramatically faster and has a dominant market share in AI chips, is trading at only about 24X forward earnings. And if we don’t use the “adjusted” numbers, and look instead at real GAAP earnings, which means that we have to assume that stock-based compensation is a real cost for these companies, then the difference is even more extreme — NVDA’s GAAP forward PE is closer to 30, thanks to the fact that their windfall in earnings this year finally washed out the impact of their stock-based compensation to a meaningful degree, but AMD, partly because they haven’t had that kind of windfall “reset” and their earnings have been slumping lately, has a GAAP forward PE closer to 80.

Will big nuclear fusion reactors depend on AMD’s chips? I dunno. Maybe. The Lawrence Livermore researchers are pretty excited about the first test machines that they’ve been using which are sort of “mini El Capitan” setups from HPE and AMD, but nothing is static in this space — they made a lot of progress with previous generations of supercomputers, too, and there will be next-wave machines that are more powerful still, maybe Quantum Computers five years from now, or maybe just the latest NVIDIA/IBM collaboration to take back the lead they lost in that market. I don’t know, and I don’t expect that there will be one particular chip or “AI booster” that’s specifically critical for nuclear fusion experiments… let alone any possible future fusion reactors that might emerge from this research in the fullness of time. Things change too fast for that.

And, of course, this is not exactly a moneymaker for AMD — like NVIDIA, they’re playing in the supercomputer space in order to improve their products, advance the science, and make sure they have a seat at the table with the government… along with the reputational and brand boost that comes from being involved with a “biggest and fastest” project. AMD’s work with HPE/Cray on El Capitan is not something that makes an impact on their income statement, and it won’t create revenue or earnings growth anytime soon — that all comes from the sale of EPYC processors to hyperscale/cloud companies (and soon the MI300 GPUs), Ryzen CPUs for PCs, Radeon GPUs for gaming rigs, etc. Those have mostly been growing, but not yet fast enough to begin to eat into NVIDIA’s dominance in gaming and data center chips.

I have great hopes for nuclear fusion, but wouldn’t buy AMD based on those hopes — that’s betting on what might be the computing platform that powers something 10-20 years from now, and I am nowhere near prescient enough to do that. For now, in the world that exists today, AMD is a solid and growing chipmaker, with some real potential, but I’d say NVIDIA still represents a better probability of being worth its current valuation. Not that NVIDIA is a lock either, of course, a bet on NVDA today depends to a large degree on the demand for their AI chips persisting at this elevated level for at least a few years before the almost-inevitable softening of that demand… but betting on NVDA to keep their position is still probably a higher-probability bet than speculating on AMD to eat into their market share dramatically in the next year or two, and at today’s valuations, with the current growth and current expectations for 2024, NVIDIA makes more sense to me. Which means I’m just talking my book, to be clear, since I still own some NVIDIA… though to be extra-clear, I’m also not that interested in risking more money on NVIDIA at this valuation.

King also teases a couple other ideas, though at somewhat lower volume… here are the clues for his second “special report”…

“AI is data intensive. Its why Big Data has been called “the most valuable resource on the planet.”

“But investors have largely overlooked data centers. Especially when it comes to AI. But AI is going to make some data centers insanely profitable in the coming years.

“Microsoft, Amazon, Google and nearly every other major company rely on these data centers. They are massive buildings. They’ve each got an average of 100,000 servers stacked on top of each other to the ceiling — all running 24/7.

“But there’s a problem with all of these servers running nonstop. They generate a massive amount of heat. So much heat that older air-cooling systems can’t keep up with new, high-powered equipment.

“That’s where this innovative firm comes into play. It invented a “liquid cooling” system that uses water, not air.

“I’m so excited about this investment, I’ve called the report The Big Data Profit Play of a Lifetime.”

Ah, another liquid-cooling play — we looked into this business a little bit last week, when Karim Rahemtulla pitched a variety of cooling companies as “AI” plays.

How does King think we should play this move to liquid cooling? He doesn’t really drop any other clues, but I suspect he’s probably pitching either Vertiv (VRT) or Supermicro (SMCI) — Vertiv is a data center service company, and part of that is designing and building liquid cooling systems, and Supermicro sells servers and modular rack systems for datacenters, including systems that allow for liquid cooling at the individual server level. I mentioned those briefly as the easiest “pure plays” on a boom in liquid cooling, as older data centers upgrade from their air cooling systems and new data centers are built to satisfy AI demand, and they are pretty reasonably situated to be worth their current valuations if this arms race for AI continues to heat up — I don’t know if they’ll gain 100% next year, as King teases, but it’s at least within the realm of possibility from this level (and they’ve both more than doubled this year, so do note that everyone else also thinks they’re hot AI stories).

And one more…

“I’ve profiled an innovative firm that has developed AI-powered robotics hardware and software that automates warehouses.

“Its clients are Target, Albertsons and Walmart. Over the next few years, Walmart — which already has its systems in 25 distribution centers — will be rolling it out to every one of its distribution centers … across the United States….

“Right now, its robotics systems are currently used in over 1,400 facilities. What’s interesting is, unlike most companies, it’s very selective with who it does business with.

“Currently, it only takes on the crème of the crop — one or two new customers per year….

“Its solutions can help companies save up to $250 million over the life of the system … it’s not surprising that it has an $11 billion backlog.

“As competition among retailers continues to be fierce — and businesses continue to try and cut costs … demand will only continue to surge.

“This stock could easily rise 300% in the next two years.”

That’s Symbotic (SYM), which was teased pretty aggressively by Luke Lango earlier this year, and which does have a huge order backlog as they fulfill their contract to retrofit dozens more Walmart distribution centers with semi-autonomous robotic systems. They’ve also partnered with Softbank on a joint venture called GreenBox, including a big equity investment, to dramatically ramp up their capacity to add more than “one or two” new customers per year, though it will be a few more quarters before we see if that’s having an impact. I do still have some call options on Symbotic, and I think they’re probably undervalued relative to their big order backlog, but will also note that lots of investors mistakenly assume this is a small company, with a market cap of only about $4 billion — it’s not, those are just the publicly traded A shares, and there is a much larger number of shares that predate their SPAC conversion a couple years ago, the company is still more than 75% owned by founder Rick Cohen, and it’s really more like a $35 billion company. Still nicely profitable, still growing fast, but nowhere near as small as some people think they are.

So there you have it, a few more growth ideas to chew on as the market tries to decide where to turn after a wildly profitable November… anything sound good to you? Think I’m too pessimistic about the timeline for “AI Energy” or the prospects for AMD to take share in AI, or have other favorite ideas in this space… or a better match for that “liquid cooling” tease? Let us know with a comment below…

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