Funding an AI Promise

This wave of AI and tech IPOs reminds me of Kickstarter.

Remember Kickstarter? Throw in a few dollars, and the sequel to a movie or book you loved gets its funding.

You weren’t buying a business. You were funding a promise and hoping it shipped. The product was the reward.

That’s how the market is treating the biggest AI companies this year.

We pay for the luxury of breaking even in 10 years.

SpaceX just ran the largest IPO in history. Anthropic has filed and could list as soon as October. OpenAI filed too, now leaning towards waiting until 2027.

These AI companies aren’t growing into profit; we’re funding them out of debt. Their next 10 years of growth are priced in.

With Kickstarter, at least you knew you weren’t investing. The danger now is that people think they are, when they’re funding a dream and hoping the profit ships.

I want to point you in the other direction.

The AI that changes your life, and by that I mean rewards your portfolio, is the AI you will never see. Here is why that matters for your money.

Paying for a pitch

Let’s start with a clean example, SpaceX.

It was priced at US$135 and ran toward a US$2.8 trillion peak, then gave back around a third of those gains within days.

It now trades near US$150, where it opened on day one.

The company lost close to US$5 billion last year. The current price isn’t a bet on what SpaceX earns. It’s a bet on what it might become in 2040.

OpenAI watched that pop and deflate, grew cautious and understands there may not be much appetite left for retail investors. Its listing was pushed back.

These are lottery tickets with a ten-year draw.

AI is paying now

You won’t see most AI’s at work as they won’t be centre stage. They’ll fill in the cracks.

Tap your card at a café and a model checks that payment in the time it takes for the light to flicker. You only notice it’s there when it detects fraud.

The phishing email that never reached you was read and binned before you logged in.

The parcel that turned up a day early took a route nobody considered.

None of this tech had a launch event. All of it runs billions of times a day. All of it pays.

You don’t know it’s there, that’s the idea

We cannot buy “fraud detection” as a stock.

What we can buy is a company that builds it, embeds it in its product and logs the revenue.

Let’s look at ASX hopeful, Pro Medicus.

It makes software radiologists use to read medical scans. Dull work, essential and built into the big US hospitals.

Its platform, Visage, is now so embedded that radiologists at the top US hospitals have started refusing to work at institutions that don’t run it. That’s not a product you build in a quarter.

Pro Medicus is rolling out AI-assisted reporting across North America, built to make radiologists faster rather than replace them.

This is what I have been writing about for the past month.

Profitable businesses aren’t AI companies; they are the payment, shipping or reporting businesses. They use AI to enhance their business, and the AI earns its keep. They are bought on today’s numbers, not a promise about 2040.

The background work is hard to value. That is why the loud pitches get the premium, and the businesses that simply work get overlooked until someone opens the books.

It’s a trap

The AI everyone is funding asks you to back its promise and wait, even though there’s already AI in the numbers earning its keep.

Next time the parcel is early, or the bill is a little lower than you expected, there it is.

That’s the AI integration that won.

Until next time.