Platformonomics TGIF #81: February 21, 2025

Platformonomics TGIF is a weekly roll-up of links, comments on those links, and perhaps a little too much tugging on my favorite threads.

News

This endeavor seems to have all the ingredients to be a YUGE success:

Dominari Holdings, Inc., formerly known as Aikido Pharma, Inc., was initially formed in 1967 as Spherix Incorporated. Since 2017, the Company has operated as a biotechnology company with a diverse portfolio of small-molecule anticancer and antiviral therapeutics in development. Over the past year, in an effort to enhance shareholder value, the Company has shifted its primary focus away from biotechnology to a new line of business in the fintech and financial services industries.

Dominari Holdings specializes in wealth management, investment banking, sales and trading, and asset management, and is seeking opportunities outside of its current business.

Kyle Wool, president of Dominari Holdings said that ADC aims to build a “robust portfolio of cutting-edge, energy-efficient data centers strategically located in key markets across the US.”

Details on locations or specifications of any planned data centers were not shared.

The AI data center bubble continues to inflate.

The hyperCAPEX companies I get (and revere for spending over $250 billion on CAPEX in 2024). They have mountains of money, actually build and operate a global computing infrastructure, and have multiple multi-billion dollar businesses that utilize that infrastructure.

Then there is the stampede to invest in data centers. There are the traditional data center companies, who aren’t seeing an AI tailwind. Beyond them, an army of financial investors have piled in who assume writing a check is sufficient to play and win at this game. My old friends in private equity and the sovereign wealth funds are pouring cash in (but not expertise or workloads, because they have none). That notoriously cautious steward of capital Softbank says it is good for $40 billion (2.5x it’s WeWork investment). Stargate’s headline number is growing an order of magnitude a month (though actual fundraising and any build-outs lag). We have the various sovereign Stargate responses: Le Stargate, K-PopGate, AmbaniGate, AdaniGate, with more no doubt to come (the scandalous -gate suffix may prove apt).

Where I struggle is who do all these AI data center investors think they’re going to lease their data center space to? There is some opportunity to lease the odd data center to the hyperCAPEX companies, but generally they prefer to build their own (here’s Satya cackling about benefiting from the data center bubble, which he charitably calls an “overbuild”. He also cackles about AGI and what you can do with just one data center in Abilene).

Who are the other customers expected to fill these data centers? GPU scarcity means early adopter enterprises will get started in the cloud, not at leased/co-location data centers. The scale dynamic also pushes people to the cloud, as you’re not going to do AI with a few servers under a desk (but it was worth a try Oracle!). There are lots of frontier model aspirants who want to train, as every AI researcher takes their shot. But the expected survival rate is plummeting, as models commoditize. The boutique GPU clouds that NVIDIA seeded in an attempt to reduce hypercloud purchasing power will have to build out a full set of cloud services to get beyond just offering training clusters. Sure, lets pretend it is plausible to catch the hyperclouds from a standing start (and a number of these neoclouds are fundamentally financial players, not technology companies).

I’m much more optimistic about the energy investment boom coincident with and catalyzed by the data center built-out, simply because that is where the scarcity is. New energy assets serve a broad range of customers beyond data centers as we atone for decades of underinvestment in the grid and power generation. A great lens on new data center projects is how much power have they secured and when will the electrons flow.

But if you really want to get ahead of the curve, the next big opportunity is going to be finding new uses for surplus data centers.

Shakedown-as-a-Service

Maybe Xitter has found its business model:

Depressingly it seems to be working as companies look for ways to bend a knee, curry favor, and avoid retribution.

What is Elon’s Tesla Strategy?

Tesla faces enormous challenges. Vehicle sales declined for the first time in 2024, even with big incentives. Early 2025 results suggest that decline is accelerating. The product line is stale in the face of immense Chinese competition. Trade barriers are rising while electric vehicle subsidies are declining.

Tesla is a meme stock trading at a ridiculous valuation to its fundamentals, and comprises the vast majority of its CEO’s net worth (who has other demands on his time). A collapse in confidence and/or a revaluation to fundamentals would be devastating.

So what is thin-skinned fractional CEO Elon Musk’s plan to salvage Tesla? The articulated hail mary is Tesla will practically overnight become a robotaxi and humanoid robot company that dwarfs its current incarnation. Details are fuzzy (he doesn’t seem to have to worry about the SEC any more):

“I see a path, I’m not saying it’s an easy path, but I see a path of Tesla being the most valuable company in the world by far. Not even close, like maybe several times more than — I mean, there is a path where Tesla is worth more than the next top five companies combined.”

What is his game plan here? Uncharacteristically deliver on product and schedule promises? Enlist the Federal Government to somehow save Tesla? What regulation/deregulation would be material? Class action lawsuits to force people to buy Teslas? (see above). Sell the company to someone? Escape to Mars? Overdose?

Share your best theories and I’ll post them next week.

(UPDATE: form removed as it is a magnet for spam)

Khakiocracy Status Check

We’re a month into the khakiocracy. Lets see how the boys in khaki are doing making the world a better place (for themselves). Definitely some wins on nuking the whole AI safety complex from orbit and letting shitcoins rip. But there are ominous developments on more substantive fronts:

Those would be Lina “all acquisitions are bad” Khan’s Merger Guidelines. Wasn’t the FTC’s stance on acquisitions the original sin that upended Silicon Valley’s political landscape?

The Moore’s Law of Energy

Moore’s Law was a great tell for whether a computing effort would face headwinds or benefit from a technological tailwind. In the energy world there is a similar tell: energy density. Batteries will never match the energy density of hydrocarbons. And nuclear is the densest.

The Unicorns Coming Home to Roost

Matt Levine chronicles the rapidly expanding universe of terms to describe the dark side of unicorn startups. As we discussed recently:

Once explicitly defined, unicorns became much less rare, as Goodhart’s Law predicts (“When a measure becomes a target, it ceases to be a good measure”).

Many games of “you get a unicorn valuation, I’ll set the terms” were played. But premature optimization of interim valuations isn’t the only thing afflicting the unicorn population.

My own terminological contribution is the “still-a-corn”:

Still-a-corn

[ still-a-kawrn ] noun

a startup still valued at a billion dollars or more even after a massive valuation haircut

They’re extraordinarily rare, like the original unicorn…

No Product, No Customers, No Revenue. Very Safe!

Too Late to Migrate Alert: Digital River

Hopefully you migrated before private equity caused Digital River to dry up. There are more such shutdowns coming, so make your migration plans a priority. Because when private equity comes amalgamating, it is time to start migrating (rhyming remains hard).

Private Equity: Yet Another Warning