The Best Thing That Can Happen to AI? Let It Get Boring.

Anthropic just filed to go public at close to a trillion-dollar valuation. OpenAI is right behind it. The AI narrative is about to collide with something it's never faced before: quarterly earnings calls.

Here's the contrarian part nobody's saying out loud: this is exactly what needs to happen for AI to matter long-term.

I watched this movie already with crypto. From inside financial services, I had a front-row seat as blockchain went from revolution to revenue model. The pattern is so clear now that I can tell you what happens next — not because I can predict the future, but because I've seen what happens when narrative meets SEC filing requirements.

When Coinbase Stopped Being Revolutionary

In April 2021, Coinbase went public on a story. The pitch wasn't "we run a trading platform." It was democratize money, dismantle the banks, rewrite the social contract of finance itself. Investors priced it like that story was inevitable truth — $86 billion market cap on day one.

Then reality showed up with a margin spreadsheet.

The stock got cut in half. Then half again. Not because crypto died, but because the market stopped paying for the narrative and started pricing the business. Revenue per user. Customer acquisition cost. Regulatory compliance expense. All the boring stuff that determines whether a company prints cash or burns it.

Today? Coinbase is a profitable, regulated, S&P 500 company that custodies over a trillion dollars in assets and reports earnings every 90 days like everyone else. It's boring. It's essential. It's still here.

That last part is the one that matters.

The IPO Is Where Stories Become Spreadsheets

Going public doesn't kill innovation. It kills the luxury of storytelling without proof.

Once Anthropic files its S-1, "democratizing intelligence" becomes a nice tagline in the investor deck, but Wall Street is going to ask about gross margin on API calls. OpenAI will have to disclose actual compute costs per query, not vibes about AGI timelines. The analysts won't care about your mission statement — they'll care whether you can defend pricing when Google and Amazon are racing you to the bottom.

This is the moment a technology stops being a pitch and starts being a business. And here's what I've learned from surviving multiple disruption cycles: the technologies that survive this transition are the ones that matter.

The ones that can't answer "how do you make money?" with specifics don't make it. The ones that do become infrastructure.

Boring Isn't the Bubble Bursting. Boring Is the Business Showing Up.

Let me give you the uncomfortable question I'm sitting with: What if boring is actually the bullish case?

Nothing kills a revolution faster than having to report earnings every quarter — and nothing builds durable infrastructure better. The narrative phase attracts capital and imagination. The boring phase builds the plumbing that actually runs the world.

Think about the internet. Pets.com died in the dot-com crash with a spectacular flameout that everyone remembers. TCP/IP — the actual protocol suite that routes every email you send and every video you stream — survived quietly and now runs the entire planet. Nobody gets excited about TCP/IP at dinner parties, but without it, your dinner party doesn't have Spotify.

Or look at the railroads. The 1840s railroad speculation bubble in Britain wiped out thousands of investors betting on dramatic returns. The rails themselves? Still moving freight 180 years later. The speculators got destroyed. The infrastructure became indispensable.

The narrative is the bubble. The plumbing is the business.

What This Means for Your Monday Morning

I work with finance leaders, auditors, CPAs — people who need to make actual decisions about AI adoption, not just have opinions about it. And here's what the IPO wave changes for practitioners:

Vendor stability becomes real. Once these companies have public shareholders and quarterly guidance, you can actually evaluate them like you'd evaluate Oracle or Salesforce. Balance sheet health, customer concentration risk, R&D spend as a percentage of revenue — all the boring metrics that tell you whether a vendor will still be returning your calls in three years.

Pricing gets rational. Right now, AI pricing feels like venture-subsidized land-grab tactics. Post-IPO, companies have to show a path to profitability that doesn't involve burning cash to buy market share forever. That means pricing that reflects actual economics, which means you can build a business case that doesn't evaporate when the VC money runs out.

Integration risk drops. Boring companies build boring things like enterprise SLAs, SOC 2 reports, and backward-compatible APIs. Revolutionary companies break things fast and ask questions later. If you're building AI into audit workflows or client-facing tools, you need the boring version.

I don't know if AI is overvalued today, and neither does anyone selling you certainty in either direction. But the destination isn't a crash — it's a boring, enormous, indispensable utility. That's not the obituary. That's the investable part.

The Question You Should Be Asking

Here's what I'm watching: Which AI companies can make the transition from story to spreadsheet?

Some won't. They'll get acquired or fade out because the unit economics never worked and the narrative was all they had. Others will do what Coinbase did — survive the repricing, build actual moats, and become the infrastructure layer that everyone forgets to appreciate until it goes down.

The companies that make it through won't be the ones with the best mission statements. They'll be the ones that can answer the boring questions with specific numbers. Cost per inference. Customer retention rate. Free cash flow. Margin trajectory.

When Anthropic and OpenAI file their S-1s, don't read the vision section. Read the risk factors and the unit economics. That's where you'll see whether this is Pets.com or TCP/IP.

What to Do Right Now

If you're evaluating AI vendors this quarter, here's what to ask:

"What happens to your pricing when you're reporting to public shareholders?" If the answer is vague or defensive, you're building on subsidized sand.

"Show me your SOC 2 Type II and your customer concentration." Boring companies have these ready. Revolutionary companies tell you they're working on it.

"What's your margin on this product, and how does it change at scale?" If they can't answer specifically, they don't know their own business model yet.

When AI finally gets boring — when the earnings calls start and the narratives get stress-tested against GAAP accounting — that's when I'll know it's real. Not because boring means small. Because boring means it survived contact with reality and became something you can build on.

The question isn't whether AI is overhyped. The question is: when the hype becomes homework, will the business still be there?

I think it will. But what do I know — I've only watched this movie three times.

What's your take? Are you waiting for AI to prove itself as a business, or are you already building on it? I'd love to hear what questions you're asking vendors right now — drop a comment or reach out directly.