In the mid-1970s, GEICO was a few months from running out of money. The stock had fallen from around sixty dollars to about two. The company had spent years writing more policies than it could afford to pay out, which is a pleasant way to grow right up until the bills arrive. A new chief executive named Jack Byrne walked in and did the least celebrated thing a leader can do: he made the company smaller. He raised prices, dropped customers, pulled out of whole states. He fixed the thing underneath before he touched the thing on top. The growth everyone wanted came later, and only because he refused to chase it first.
I keep that story close this week, because the loudest signal in the market right now is the opposite instinct.
The headline doing the rounds is that the most-watched company in artificial intelligence has not held a single pre-IPO investor meeting and has not set a timeline for going public. Read one way, that is a non-event — a company declining to do a thing it was never obligated to do. Read another way, it is the most interesting thing the company has said all quarter. The entire market is leaning forward, asking for a date, and the answer is a shrug. No meetings. No timeline. Not yet.
The market hates “not yet.” A date is a story you can trade. “Not yet” is just the truth, and the truth doesn’t move quickly enough to be exciting.
The demo is not the deployment
Somewhere in the same week, there’s a quiet recurring conversation — a weekly one — about bringing agentic AI “to production.” Notice that it has to be a recurring conversation. Nobody schedules a standing weekly meeting about a solved problem. The phrase “to production” is doing enormous work there. It is the whole distance between a thing that demos beautifully and a thing you can actually leave running while you sleep.
A demo is a performance of readiness. Production is the absence of an audience. One is built to be watched; the other is built to be forgotten about, which is much harder, because forgetting about something is the highest compliment you can pay it. It means it works.
This is the same fault line running under the IPO non-news and the GEICO story. There is the version of a thing that looks ready, and there is the version that is ready, and the gap between them is where most of the actual work lives — and where almost none of the attention does. Attention is drawn to the announcement. The work happens in the silence before it.
Scarcity is a feature of value, not a flaw
A fourth signal floated past me phrased as a question: how do you ration a consumer right? It reads like a finance puzzle, and it is, but underneath it is something simpler. The moment something is worth having, you have to decide who gets it and when. Rationing is just the honest admission that not everything can be available to everyone at once.
A company that won’t name an IPO date is rationing certainty. A team that keeps an agent in testing instead of shipping it is rationing trust — releasing it only when it has earned the release. Byrne rationed GEICO’s own growth, deliberately, when growth was the only thing the market wanted to buy. In each case the restraint looks like weakness to the person who wants the thing now, and looks like discipline to the person who has to live with the thing later.
We have built an economy that treats availability as the goal — everything, instantly, to everyone. But the things that hold their value tend to be the ones that made you wait. Not as a marketing trick. As a side effect of actually being finished.
Profit is a by-product, not a plan
There’s an idea I keep circling: that a great customer experience produces profit as a happy by-product, and that aiming straight at the profit puts you directly at war with the customer you’re supposed to serve. Aim at the experience and the money follows. Aim at the money and you start writing policies you can’t pay out — which is, precisely, what nearly killed GEICO.
The same logic scales up to the whole AI moment. Aim at the valuation, and you set a date, hold the meetings, ship the half-ready agent, perform the inevitability. Aim at the thing actually working — the underwriting sound, the agent durable, the experience worth having — and the valuation becomes a by-product you don’t have to chant for.
The market keeps asking these companies when they’ll be ready, and treating the absence of a date as a problem to be solved. It has the question backwards. The companies that survive their own hype are usually the ones disciplined enough to say “not yet” while everyone else is selling “now.” Byrne’s “not yet” saved a company everyone had already written off. We remember it as a turnaround. At the time it just looked like a man refusing to perform.
The hardest thing to sell is patience, because it has no launch date. But it is the only thing on this list that has ever reliably compounded. Everything else — the timeline, the demo, the announcement — is just the appearance of readiness, rented by the day. Real readiness doesn’t announce itself. It just quietly keeps working after the room has emptied out.
Leave a Reply