AI’s “Big, Dirty Secret”

Buried in the middle of James Altucher’s Musk Millionaire presentation is the part we found most interesting, and it has nothing to do with the briefcase or the 1,806,000-new-millionaires headline. It’s the claim that artificial intelligence has an energy problem that terrestrial infrastructure cannot solve — and that the tech industry’s biggest names have all said so on the record.

Altucher assembles the quotes. Jensen Huang, CEO of NVIDIA: “Every data center of the future will be power limited.” Sam Altman, head of one of the world’s biggest AI companies, telling Congress it’s “hard to overstate how important energy is to the future of AI.” Mark Zuckerberg admitting Meta would “build its AI far bigger, if he could just get his hands on enough electricity.”

When the people spending the most on AI compute all say the same thing about power, it’s worth listening.

The 840,000-Homes Math

The presentation’s anchor statistic: just one data center can use as much power as 840,000 homes — the entire city of Indianapolis, for a single building complex. And unlocking AI’s full potential, the argument goes, requires thousands more data centers of that size.

The downstream effects the presentation cites are concrete and mostly verifiable in news coverage:

  • Data centers are “one of the most debated topics this year,” disrupting communities with noise pollution and tanking housing values
  • Electric costs in affected areas have “increased by more than triple”
  • Lawmakers across 30 states have filed over 300 bills to slow data center projects
  • Local communities have killed $98 billion in proposed builds
  • Some California cities are banning them outright

Whether or not you buy the investment pitch built on top of it, this is a real dynamic. The AI buildout is colliding with a power grid that was not designed for it, and the friction is political as much as technical. The presentation frames the stakes starkly: if the trillion-dollar AI industry can’t be powered, “the full potential of AI would always be a pipedream” — cancer cures, full automation, and everything else the technology promises.

The Ground-Level Scramble

The most useful part of the presentation’s energy section is the survey of how each tech giant is responding, because it shows how uniformly reactive the industry has been:

  • Amazon is investing $20 billion in Pennsylvania for AI, including nuclear
  • Microsoft is trying to restart old nuclear plants to power AI infrastructure
  • Meta is funding seven gas plants to power its biggest data center

These are the moves of companies that were, in Altucher’s words, “caught flat footed, stuck on the ground, scrambling for every last watt of power they could dig up.” Gas and nuclear are multi-year, multi-billion-dollar commitments being made under competitive pressure — which tells you how serious the constraint is.

Elon’s Different Answer

The pivot of the Musk Millionaire thesis is that Elon Musk looked at the same problem and chose a different coordinate system. Instead of buying power on Earth, the FCC filing at the center of the presentation describes one million solar-powered AI satellites, with solar panels working “up to eight times better” in orbit, free cooling at 250 degrees below zero, and no grid, permitting, or community opposition to fight. We cover that plan in detail in our FCC filing article.

The energy argument is what makes the space plan more than a sci-fi pitch. Every constraint the terrestrial giants are struggling with — power generation, cooling, permitting, community backlash — is either free or nonexistent in orbit. The question is whether launch costs and orbital data transmission economics can beat a gas plant, and that’s genuinely unresolved.

Where This Fits in the Bigger Picture

The AI energy bottleneck has become fertile ground for newsletter analysts, each taking a different angle on the same underlying math. Luke Lango’s XPanse research and Jason Bodner’s Accelerated AI presentation both treat AI infrastructure constraints as the defining investment story of the decade — Bodner’s version puts copper and photonics at the center, Altucher’s puts electricity and orbit. We’ve also covered the nuclear angle in our small modular reactor stocks article, which is the other mainstream “solve AI’s power problem” trade.

The differences among these pitches are instructive. Nuclear and gas plays monetize the bottleneck on Earth. Altucher’s space thesis bets on the bottleneck being routed around entirely. Both can’t fully be right at once, and the outcome likely involves some of each — which is exactly why understanding the underlying energy math, not just the stock picks, is the valuable part.

For the full breakdown of the Musk Millionaire offer, the free Intel ticker, and the sub-$100 “AI MasterKey” supplier stock, see our complete review. If the power-generation investment angle is what interests you, we go deeper in our AI data center power stocks article.

NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.