RISE A Real Estate Company

The Capital Bet and Its Uncomfortable Math

Think AI will take over the world? Maybe. But I think the larger issue is the economic inflection point no one’s talking about. Given the staggering amount of capital being invested in AI, the question should be: when will it be profitable?

The ROI is troubling. AI revenue remains concentrated in a relatively small number of use cases, like coding assistants, enterprise automation and API consumption. The monetizable value generated doesn't remotely justify current infrastructure spend. The numbers are staggering. “Big Tech” players are collectively spending hundreds of billions annually on AI infrastructure: data centers, GPUs, energy and talent. OpenAI alone is burning through capital at a pace that might’ve been deemed reckless in previous tech eras.

Beneath the surface, three structural risks are quietly compounding.

● Veil of "Picks and Shovels" - Nvidia and hyperscalers are printing money selling the buildout, masking whether the end-use applications are actually profitable.

● Concentration Risk - With five to six companies absorbing most of the capital, when the correction hits, the downstream effects on the broader tech economy will be severe.

● Energy Infrastructure Debt - The power grid investment needed may be the long-tail liability nobody's properly pricing into the equation.

The Long Wait for Investors

Some of this capital is creating durable infrastructure that will be amortized over decades, similar to the broadband overbuild of the late '90s. Sure, the overbuilders mostly died out — but we all got cheap internet. This time, the difference may lie in the pace and depth of the bets.

“Given the staggering amount of capital being invested in AI, the question should be: when will it be profitable.”

Which brings us to the trillion-dollar question: how long will investors wait for an ROI? The historical patience benchmarks are not encouraging:

● Dot-com investors waited 3-5 years.

● The AI spending wave started around 2022-2023; we’re already 3-4 years in.

The ones spending massive sums aren’t scrappy startups. They’re established tech giants like Microsoft, Google and Meta. They have other revenue streams subsidizing their bets. This significantly extends the runway, reducing the likelihood of a crash. The hyperscaler argument cuts both ways. Shareholders still expect returns, but activist pressure tends to build below the surface before it breaks into the public discourse.

The real tripwire won’t be the Magnificent 7. It will be the second and third-tier players that bought heavily into the promise of AI transformation but failed to realize productivity gains at the scale sold to their CFOs. When these customers begin pulling back on AI spend, the hyperscalers will feel the lost revenue. That downturn will evaporate investor patience fast. I think we have 18-36 months before this revenue pressure becomes publicly undeniable.

The Quiet Failure on the Demand Side

There’s another looming problem: enterprises and SMBs can’t seem to find a path to return on investment, and that's the dirty little secret nobody on the vendor side wants to talk about, and this is why:

● Pilot Purgatory: endless proofs of concept that never scale to production.

● Productivity Theater: coding agents save developers some time, but not enough to justify the broader transformation narrative

● Cost Substitution Without New Revenue: replacing headcount with automation doesn’t grow the top line.

For SMBs, the problem is even worse. The tooling assumes a level of data maturity, IT sophistication and change management capacity that most SMBs simply do not have. The gap between what AI can theoretically do and what a 200-person company can actually implement is yawning.

On the vendor side, accountability is largely absent. AI vendors are selling capability, not outcomes. There's no SLA on ROI. When the renewal conversation comes around, and a CFO asks, "What did we actually get for our money?" the justification is thin at best.

The underlying structural friction that most organizations haven't solved is data hygiene, and you can't AI your way out of dirty data, fragmented systems, or unclear processes. Even if your organization is asking the right questions, they are still being sold a solution in search of a problem.

The ROI Reckoning Is Here

Investor patience has a sell-by date. We're roughly 3 years into serious deployment capital. While the Magnificent 7 have longer runways than startups, enterprise customer pullback will be the tripwire that makes investor pressure undeniable. Neither enterprises nor SMBs have found a credible path to ROI. Between pilot purgatory, immature infrastructure and vendors selling capability instead of outcomes, the demand side of the equation is quietly failing to validate the supply side's enormous bets.

● Capital investment in AI is likely unrecoverable at the current scale.

● The space between infrastructure spend and monetizable outcomes isn’t closing.

● Buildout is accelerating faster than viable revenue models.

We may be watching the construction of the most expensive bridge to nowhere in the history of capital markets or, at a minimum, one that arrives about a decade too early for the traffic it was built to carry.