Ray Dalio writes about the Dutch East India Company the way you’d write about a species discovering fire. “The Dutch created the world’s first mega-corporation,” he notes in Principles for Dealing With the Changing World Order, one that “accounted for about one-half of all world trade.” What made it possible wasn’t the ships. It was the new financial instruments invented to fund them — capitalism itself, assembled on the fly to serve ambition that had outgrown existing tools.
That sentence has been sitting differently this week.
SpaceX and OpenAI are reportedly preparing for public offerings that could break records for size. The analyst commentary arriving alongside this news is predictable: mega-IPOs signal market tops, liquidity seeking an exit, insiders cashing out before the tide turns. It’s a reasonable frame. History has some data behind it. But it’s also the most surface-level read available, and surface reads tend to miss what’s actually happening in the infrastructure underneath.
What’s actually happening is this: the same week those IPO headlines dropped, Amazon announced AgentCore Payments — a system designed to let autonomous AI agents transact on their own, without waiting for a human to approve each purchase. Anthropic is in talks with Microsoft to use their in-house AI chip, the MAIA 200, as part of a deeper infrastructure partnership. Circle is expanding USDC’s reach across more settlement layers. MoonPay launched an institutional trading platform, building out the rails for serious money — and eventually machine money — to move through crypto markets with more efficiency.
These aren’t separate stories. They’re different views of the same thing being assembled.
The Dutch East India Company needed capitalism to exist before it could exist. The next generation of large-scale economic actors — AI agents, not companies in the traditional sense — need payment infrastructure, compute at the chip level, and programmable money before they can operate at scale. That’s what this week looks like when you step back far enough.
What a Market Top Doesn’t Explain
The “mega-IPO equals market top” argument is about valuation timing. It’s asking: are we near the peak of what the market will pay for these assets? That’s a real question. But it’s a different question from: are these companies building infrastructure that will persist and compound regardless of where equity prices go in the next eighteen months?
$NVDA didn’t stop being important when it pulled back in 2022. The underlying architecture it was supplying didn’t pause. The same logic applies here. Whether $AMZN’s AgentCore Payments is priced correctly into the stock is separate from whether autonomous agent commerce becomes a real economic layer. One question is about timing. The other is about direction.
The systems thinker’s real job is watching for the variable that doesn’t show up in the headline metric — the place where profits drop even when output stays consistent, because something structural shifted that the obvious dashboards don’t measure. Right now, the obvious dashboards are measuring IPO size and valuation multiples. The structural shift is in who controls the compute, who owns the payment rails, and who sets the settlement standards for transactions that no human initiates.
The Chip Layer Is Where This Gets Decided
The Anthropic-Microsoft chip deal is easy to overlook next to a potential record-breaking float. Don’t overlook it.
Whoever controls the chip layer controls the cost structure of inference, which controls the economics of every AI product built on top. $MSFT designing its own silicon and licensing it to Anthropic isn’t just a cost play. It’s vertical integration of a kind that changes the leverage relationships across the whole stack. The Dutch East India Company eventually needed its own army because the system it built became too complex to operate through external parties alone. The AI industry is reaching a version of that same inflection — the leading labs are moving toward owning their compute rather than renting it.
That’s a durable structural story. It runs underneath the IPO timing question, independent of whether the float happens this year or next, whether markets are at a top or mid-cycle.
What the Crypto Rails Are Actually Building For
MoonPay building institutional infrastructure and Circle expanding USDC settlement look, on the surface, like crypto firms pursuing mainstream legitimacy. That framing is correct but incomplete. The longer arc is that programmable money — money that can be held, moved, and deployed by software without human intervention — is prerequisite infrastructure for the agent economy Amazon is trying to build with AgentCore.
Human payment rails have friction by design. Approvals, compliance checks, human signatories. Those friction points exist because the actors were assumed to be human. When the actor is an AI agent executing a workflow at machine speed, the friction becomes a bottleneck. Crypto settlement layers, stablecoin infrastructure, programmable wallets — they’re not just fintech. They’re the payment primitive the agent economy needs.
The Dutch invented the instruments before anyone fully understood what they were building. The company came later, larger than anyone anticipated. That sequence — instruments first, then the entities that use them — is worth keeping in mind right now. The payment rails, the chips, the settlement layers. The mega-corporations that use them are still getting their paperwork in order.
The IPO isn’t the event. It’s the moment we start tracking something that’s already been running for a while.
Leave a Reply