AI, Capital Cycles, and What Happens Next
By Don Hoang
There’s a version of the AI story that gets told loudly—record funding rounds, inflated valuations, benchmark-defining performance. But underneath all that is a quieter, more enduring truth: we are in the early stages of a modern Industrial Revolution, and it’s moving faster than most people can track.
AI will reshape work, capital markets, and society. It will automate labor at scale, create new markets, and—importantly for investors—generate a cohort of companies that become generational winners.
That’s the long arc.
But we’re not in year one anymore. And cycles matter.
1. AI is a true industrial shift
AI is not just a new tool; it’s a foundational shift in how value is created. Not unlike steam power or electricity, this wave will impact every knowledge-based industry, from compliance to customer service to creative production.
It’s already happening:
Tier-1 law firms are running hundreds of client matters through LLM-powered interfaces.
Fintechs are compressing workflows from days to minutes using embedded AI agents.
Enterprises are rebuilding internal systems around model-native architectures.
This shift won’t be linear, but the direction is set. I believe we’ll see 25 to 200 companies emerge over the next decade that are multi-decade compounding machines—defining new categories, building high-margin businesses, and becoming systemically important.
2. We’ve moved from early deployment to frenzy
If 2022–2024 was about deployment and infrastructure—building the foundation and early application layers—then 2025 marks the turn into the frenzy phase.
Capital is pouring into the space. Valuations are elevated. Suppliers are investing downstream in customers. There’s increasing circularity in the ecosystem—infra providers backing usage layers to drive compute demand, enterprises deploying budget just to “be in the game,” and funds stacking exposure across the stack.
Expectations are high. In some cases, unreasonably high.
At this point, many investors aren’t underwriting fundamentals—they’re chasing velocity and story. That’s always a late-stage signal.
3. Overbuild is next—and then a correction
If this continues (and it will), the next leg of the cycle is straightforward: overbuild.
Too many companies will get funded to do too many things—model training, orchestration, workflow apps, co-pilots, agents, vertical SaaS, foundation infra. They’ll all chase talent, GPU supply, and enterprise pilots. For a while, that will work.
But eventually:
– ROI will come under pressure.
– Usage won’t scale as expected.
– The infra costs will eat into gross margins.
– Renewal rates will disappoint.
Most companies won’t hit their numbers. Investor confidence will crack.
Capital will tighten. And the bubble will pop—just like every other platform cycle.
4. Timing the break: late 2026–early 2027
It’s impossible to time a market perfectly. But directionally, here’s how I see it:
The current capital easing cycle likely ends mid-2026.
Most of today’s AI leaders raised in 2023–2025 on rich terms.
By 2H26 to 1H27, those same companies will need follow-on capital, prove their economics, or justify $100M+ valuations.
This is the window when the capital markets will test the thesis.
Some will raise again. Most won’t. The disconnect between expectation and delivery will become too large to ignore.
5. But the winners will hold—and access will improve
This correction will be painful, but it won’t be 2000 or 2022 all over again.
Why?
Because the underlying businesses—at least the top decile—are stronger.
They have:
Real customers
Rapid revenue growth
Infra leverage
Structural gross margins
A clear path to market leadership
Think:
Facebook post-2008 – massive growth amid market collapse
SpaceX in 2022–23 – outperforming while capital retreated
Databricks, Stripe, Snowflake – enduring platform businesses that strengthened through cycles
These companies will get cheaper, not broken. And access—which today is nearly impossible—will open up.
6. Position now to win later
The opportunity is not in guessing the exact week the cycle turns. It’s in building conviction now—before others do.
If you’re deploying capital in 2026–2028, this is your window.
If you’re advising LPs or raising a new vehicle, this is when differentiation matters.
If you’re an operator, this is when you want to be at the right company—not the most talked-about one.
Not to chase this phase, but to be positioned for the one after.
Because when the dust settles, the next cohort of foundational companies will emerge.
And the people who showed up early—with conviction, not FOMO—will be the ones who own it.
