SpaceX just completed the largest IPO in history, then gave back almost a third of its value within days. That swing is the headline, not the signal. The signal is quieter: The most patient capital in the world has already decided who will own the infrastructure, before the rest of us can even see the walls.
Consider the chess pieces moving right now. Anthropic is targeting a near-trillion-dollar valuation. Alphabet recently raised $85 billion in equity to fund its AI compute infrastructure — with Berkshire Hathaway anchoring $10 billion of it. Even through the post-IPO sell-off, SpaceX continues to cement its position as the ultimate landlord of an outer infrastructure economy.
Read all this as a speculative frenzy and you will draw exactly the wrong conclusion. Look instead at who is moving, and why.
Berkshire Hathaway does not chase momentum. Under Greg Abel, its discipline remains unchanged: buy infrastructure once the survivors are obvious and cheap. So when Berkshire commits $10 billion to anchor an equity raise earmarked for AI infrastructure during the buildout rather than after it, it isn't enthusiasm. It is a judgement. The most patient capital in the world has concluded the winners are already identifiable, and it’s willing to break a decades-old aversion to tech to back them.
While many still like to create headlines of red numbers and talk bubbles, the signal says infrastructure consolidation, financed by the people who fund railroads, not features.
The Macro Reality: Landlords and Tenants
At the macro level, this new economy is binary: You are either the landlord writing the lease or the tenant paying whatever the landlord decides to charge.
Most leaders assume they sit safely in a third category. They do not. A handful of AI enablers captured the overwhelming share of recent market profit growth. Strip them out and the rest of the market is remarkably close to flat.
The reason is uncomfortable: Most companies are buying AI while clinging to legacy processes and legacy tools. They are spending millions to rent capability while changing nothing about the terms on which they operate. That is the posture of a tenant. More rent, same building, someone else’s name on the door.
But becoming a landlord is not a switch you flip. You cannot claim it by putting "infrastructure" in a press release. Landlords win by accumulating a form of dependence that others eventually have no choice but to rely on.
Google didn't just build a search engine; it accumulated so much usage and data that the rest of the web had to rent its ecosystem.
Constellation Energy didn't build new tech. It recognized that its massive nuclear fleet held the one thing the AI buildout desperately needed: 24/7 power. By locking Microsoft into a 20-year contract to restart Three Mile Island, it was instantly re-rated from a sleepy utility to an infrastructure landlord.
It recognized the dependence before the market did. That is the exact market read this moment demands of every corporate leader.
The Micro Playbook: The Three-Layer AI Strategy
Faced with a market run by powerful landlords, how do you position your own business? You cannot own all the infrastructure, but you cannot afford to blindly rent your entire future either.
To locate yourself honestly before the market locates you, you must deliberately map your operations into a three-layer AI strategy:
- The Core (what you protect and compound): This is your uncaptured intellectual property, proprietary data, or structural workflow advantage. This is the dependence that is genuinely yours to own. If you give this away to a LLM vendor, you are volunteering to become a tenant of your own IP.
- The Peer Layer (what you partner and pace): This is where you trade usage for efficiency to keep pace with the market. Here, strategic partnership beats outright ownership, provided you know exactly what value you are getting back for what you put in.
- The Outer Layer (what you outsource): Everything here is interchangeable and commoditized. The only honest question is “cheapest or best?” Renting here is entirely rational.
The mistake most organizations make is not financial, it is failing to be honest about which layer a capability actually sits in.
Take a much-discussed alliance: Microsoft’s partnership with OpenAI. It cleverly lifted Azure’s growth at a time where the main number being scrutinized by the market was cloud growth. But when OpenAI later prioritized access to premium compute terms to SoftBank, Oracle, and AWS, the reality confirmed that OpenAI’s capabilities are not in Microsoft’s Core but sit in its outer layers. A perfectly viable strategic choice — but a dangerous one if leadership misdiagnoses it as an exclusive asset.
The Landlord’s Question
Without drawing these lines deliberately, your Core will erode into the Outer Layer by default. You will wake up one day renting your future from a landlord who understood the value of your dependence earlier than you did.
When the most disciplined money in the world treats compute and connectivity as a utility, “wait and see” stops being prudence. It becomes a decision to rent.
Before your next AI investment review, stop asking “how much” and “how fast.” Ask the landlord's questions instead:
- What are we accumulating that others will eventually have to depend on?
- Are we compounding that advantage or giving it away to our vendors?
- Are we deploying capital to take part on someone else’s terms or to set our own?
The landlords of this economy will not win by spending the most or announcing the loudest. They see early what the world will depend on, and they secure it while it still takes conviction rather than confirmation.
Clarity comes first. Real momentum begins before the first move.
