Listen to the audio version of this article (generated by AI).
In 2009, an Israeli venture capitalist watched a video in total disbelief.
“This has to be fake,” he thought, as a person moved in front of a camera and a digital skeleton followed along on screen, mirroring the person’s movements in real time.
But Eden Shochat was getting a real demonstration from PrimeSense.
PrimeSense’s technology went into Microsoft’s (MSFT) Kinect gaming system. In 2013, Apple (AAPL) bought the company for a reported $350 million.
One of its founders, Aviad Maizels, eventually started another business, Q.ai. This time, his team worked on technology involving audio, machine learning, and subtle facial movements.
Shochat got another look at a prototype that seemed almost too ambitious to believe.
This time, he invested.
In January, Apple acquired Q.ai for a . By the time most investors heard about the company, Apple had already bought it.
I keep thinking about that distinction as the AI industry debates whether to slow the development of its most powerful models.
On Wall Street, investors are asking what a longer wait for the next breakthrough could mean for AI companies’ growth, and what their stocks are worth today.
But many smaller companies are working on a different problem: How do we turn the AI we already have into something customers will pay to use?
That could mean helping a factory spot defective parts or teaching a robot to perform a useful task.
Those businesses don’t necessarily need a more powerful AI model to keep growing. They need to make existing technology reliable, affordable, and useful enough to win customers.
And that’s the opportunity private markets can offer during a public-market slowdown: a chance to own companies whose next stage of growth comes from solving those practical problems, even while enthusiasm for publicly traded AI stocks cools.
A private startup can keep winning customers, increasing revenue, and becoming more attractive to a potential buyer without having its shares repriced every time sentiment shifts on Wall Street.
That doesn’t make it immune to a downturn, but it creates another way to participate in AI’s growth… through business progress that can continue even when the stock-market rally doesn’t.
Pacing the Frontier Leaves Plenty of Work to Do
On Sept. 12, Anthropic CEO Dario Amodei called for a more .
I recently explored whether this represents his “Oppenheimer moment.” But for investors, the useful question extends beyond the historical comparison.
What, exactly, would be slowing down?
There is an enormous amount of work between demonstrating an impressive capability and making it useful every day. Someone has to connect the software to a customer’s systems, make it reliable, reduce its cost, and teach employees to use it. Most importantly, someone has to prove that it saves more money than it consumes.
That work creates businesses. And it can continue even while investors become less enthusiastic about AI.
Consider a startup that helps a factory spot defective parts. Its next year of growth might come from installing cameras on more production lines, improving accuracy, and winning a second customer. None of those achievements requires the entire AI industry to break a new intelligence record.
Now, imagine that company reaches those milestones during a selloff in AI stocks.
A publicly traded business could announce similar progress and still see its shares fall. Investors might be reacting to higher interest rates, disappointing earnings elsewhere, or a headline that changes their expectations for the entire sector.
A private company generally doesn’t face that minute-by-minute public accounting. Its next financing, a share transaction, or a buyout can provide a new reference point for its value. Between those events, its founders can keep building, signing customers, and improving the product.
That’s what interests me about private markets during a public-market slowdown: the opportunity to own a business that is making measurable progress while the broader AI story is being repriced.
Now, I’m not suggesting you buy private companies simply to stop seeing red numbers on a screen.
I want to find businesses whose next milestone depends on serving a customer, with progress I can evaluate through installations, repeat orders, and improving economics.
If those businesses can keep building value while public markets struggle, investing before they reach the stock market could offer an opportunity worth considering.
That’s where I’m looking.
The Buyers Still Have Problems to Solve
A longer wait for the next frontier model doesn’t eliminate the need for better interfaces, more reliable automation, or cheaper ways to deploy the systems already built. In some cases, it could make those improvements more valuable.
The giants can develop those improvements themselves. They can partner with specialists. Or they can buy a company that has already done the difficult work.
Alphabet (GOOGL) bought Android and YouTube. Meta (META) bought Instagram. Apple bought PrimeSense and Q.ai.
Different technologies across different eras, but a familiar business decision: acquiring an existing capability can be faster than recreating it.
That’s why I watch the companies receiving the buyout checks as closely as those writing them. A young company’s first product may give us only a partial view of what its technology (and the team behind it) could eventually become.
The challenge is recognizing that potential early, when the business is still taking shape.
Even experienced investors miss it. Bessemer Venture Partners maintains an “anti-portfolio” of businesses it passed on, including Google, eBay (EBAY), PayPal (PYPL), and FedEx (FDX).
Those missed opportunities are a reminder of how difficult it is to evaluate a company before its success becomes obvious. You have to assess what the founders have built, what remains unproved, and whether they have a credible path forward.
That’s where my People, Product, and Timing (or PPT) framework comes in.
People comes first because early companies rarely develop exactly as planned. I want founders who can adapt, recruit talented colleagues, and use their capital well.
Q.ai’s Maizels had already built a company Apple wanted to own. That gave investors concrete experience to investigate: what he had built, how he had executed, and whether those strengths could carry into another business. It didn’t guarantee another sale.
Product means asking what problem the company solves and whether customers care enough to pay. An impressive demonstration is a starting point. I want to understand whether the technology works repeatedly, whether the economics make sense, and how useful it could become beyond its initial application.
That last question matters when considering a potential acquisition. A technology serving one narrow market today could address a problem inside a much larger company.
Timing means understanding why the opportunity exists now. Has the technology become affordable? Are customers ready to use it? Does the company have enough cash to reach its next meaningful milestone?
Those questions help me evaluate whether a startup can build something valuable. Then I examine the valuation and investment terms to determine whether that potential could translate into a worthwhile return.
Finding opportunities like that means evaluating the people, the product, and the timing while the business is still private, and being disciplined about what you pay.
One Robotics Company Brought This Into Focus
That framework led me to the private robotics opportunity I discuss at .
Its early work involved something wonderfully ordinary: making coffee.
Think about what that requires from a machine. It has to recognize objects, move precisely, handle equipment, and repeat a sequence reliably in a real environment.
The larger opportunity is in the system that teaches the robot how to do those things. That’s what interested me about the company’s effort to turn its operating experience into a broader robot-training platform.
If that technology can help other businesses train useful machines more efficiently, its potential extends well beyond the coffee counter.
That’s why I’ve described it as a potential “.” The company still has to prove it can build a successful platform business. But helping businesses train robots remains a valuable problem to solve, even if frontier AI development becomes more deliberate.
In the presentation, I explain the founders’ backgrounds, the technology, and why I recommended the company. I also address the challenges ahead: scaling hardware, competing with well-funded rivals, and turning its training platform into a successful licensing business.
This is the kind of research I built to provide. Members receive detailed Opportunity Memos, guidance on getting started, and ongoing research as we build a portfolio of private opportunities over time.
The goal is to help readers evaluate promising companies while they’re still private — before an IPO or acquisition changes the opportunity.
But you don’t have to purchase a membership to hear this recommendation.
We’ve , and it closes Monday, Sept. 21, at midnight.
If you missed the original event, give yourself time to watch, understand the business, and review the offering materials. Then decide whether the investment belongs in the speculative portion of your portfolio.
