The U.S. just made one of its biggest nuclear announcements of the year.
Most people will read the headline and move on.
I think the market is underestimating what the DOE is actually trying to build.
🧵
For investors, stories like this rarely create overnight moves.
But they reinforce something more valuable:
Companies deeply embedded in long-cycle defense programs often benefit from years of predictable revenue, recurring upgrades, and high barriers to entry.
This aligns with a broader trend across U.S. defense:
• Long-term production contracts
• Supply chain resilience
• Industrial capacity
• Sustainment over one-off procurements
Governments increasingly view manufacturing capability itself as a strategic asset.
Preliminary approval is the easy part to overlook.
The real signal is that regulators are now working through the framework for commercial D2D deployment rather than debating whether it should exist.
If Japan moves from testing to licensing, ASTS gains another proof point that its model can scale internationally—not just in the US.
The biggest investment opportunities often emerge before procurement data shows up in quarterly earnings.
Watch how governments change incentives—not just how much they spend.
That’s where tomorrow’s winners often appear first.
Everyone is watching how much the Pentagon is spending.
Almost nobody is watching how it’s changing the way it buys technology.
That may end up being the bigger story.
🧵
This also reinforces a broader trend we’ve been tracking:
Governments aren’t just increasing defense budgets.
They’re trying to compress the time from innovation → contract → deployment.
In an era where technology evolves in months rather than years, speed itself becomes a strategic advantage.
@OpenAI@paulg Price cuts like this put pressure on every frontier lab. If capability improvements start arriving alongside rapidly falling inference costs, differentiation has to come from somewhere other than raw model performance.
@Vertiv The interesting shift is that cooling is becoming a software problem as much as a mechanical one. As rack densities climb, the companies with the best control algorithms—not just the biggest cooling systems—could end up extracting the most compute per megawatt.
@StockSavvyShay Regulatory approval is only half the moat. The harder challenge is proving you can operate safely, reliably and profitably at scale. That’s where the real winners will separate themselves.
@CoreWeave AI is quietly becoming part of the financial system’s critical infrastructure. Once trading firms optimize around a specific compute platform, switching costs become much higher than they appear.