AI gold rush? More like a billionaire poker game.
I’ve covered Silicon Valley since the dial-up days, watched fortunes made and torched on every hype cycle from Web 1.0 to crypto winter. And now? Family offices—those secretive stashes of old money—are shoving VCs aside to cram cash straight into AI startups. No middlemen. Just cap tables and crossed fingers.
Companies drag their feet on IPOs longer than ever. Private markets? Dominated by AI darlings. Mitch Stein from Arena Private Wealth nailed it on TechCrunch’s Equity podcast:
“A lot of money is being made well before companies go public, and right now the private markets are dominated by a lot of these AI names. The family offices who are allocating [directly into AI startups] are right on.”
Arena just co-led a $230 million round in Positron, an AI chip hopeful. Board seat included. They’re not passive anymore—no sir. Stein calls it becoming “active participants.” Sounds noble. But let’s cut the spin: it’s FOMO on steroids.
Why Family Offices Can’t Sit This Out?
Urgency’s thick as fog in the Valley. Ari Schottenstein, Arena’s alternatives head, puts it plain:
“The world’s AI infrastructure is being built now, so you’re either going to get in early and have an opportunity to do more primary investing…and really build a portfolio, or you’re going to miss it and be taking random bets.”
Stein? Blunter still: Your biggest risk ain’t the downside—it’s missing the upside. Numbers back the panic. February alone: 41 direct startup deals by family offices, almost all AI. Laurene Powell Jobs into World Labs. Azim Premji’s crew at Runway. Eric Schmidt’s Hillspire grabbing Goodfire. BNY Wealth says 83% of these offices peg AI as top priority next five years. Over half? Already exposed.
Here’s my unique twist, one you won’t find in the original fluff: this mirrors 1999’s dot-com madness, when family money flooded Pets.com and Webvan pre-IPO. Back then, everyone thought “internet infrastructure” was the sure thing. Billions vaporized when the bubble popped. Today’s AI infrastructure bet? Same vibe. Chips, models, data centers—sure, vital. But valuations are moonshots already. Positron’s got Oracle as a customer, Arm on the cap table. Impressive. Yet one hyperscaler flop, and poof—your “active participation” is a boardroom tombstone.
Some offices aren’t just betting. They’re building.
Jeff Bezos plays CEO at his $30 billion robotics outfit—raised $6.2 billion last year. Smaller fry? Tyson Tuttle, ex-Silicon Labs boss (flipped to TI for $7.5 billion), spins up Circuit with AI for manufacturing. His family office drops $5 million in a $30 million angel round.
Not all are ex-founders, though. Arena’s institutional suits swear by due diligence. “Slow yes, lots of nos,” Schottenstein boasts. They vet tech with experts, eyeball cap tables—Arm’s in? Tech must be legit. Oracle buying? Not just Nvidia/AMD wannabes.
But here’s the cynical gut check: they’re doing few deals a year. No portfolio spray-and-pray. All eggs, one AI inference chip basket. Stein admits: stakes sky-high, no failure modeled in. Concentrated risk? That’s code for “we could get nuked.”
Is Direct AI Investing Smarter Than VCs?
VCs spread bets like peanut butter. Family offices? Laser-focused. Arena’s Positron play: their sole AI chip bet. Win big or wipe out. Smart? Depends. VCs have seen more flameouts—pattern recognition’s their edge. These wealth advisors? New to the rodeo, chasing “infrastructure now” narratives peddled by founders with pitch decks thicker than my expense reports.
And the PR spin—oh boy. “Active participants.” Please. It’s gamblers with advisors, not visionaries. Remember when private equity swore they’d “transform” industries post-2008? Many just juiced returns till the music stopped. Prediction: half these direct deals sour by 2027. AI’s real, but early bets on chiplets and inference plays? Riskier than a Tinder date in 2024.
Numbers scream frenzy. Fewer IPOs mean locked-in liquidity deserts. Family offices, flush from decades of compounding, smell blood—or gold. But who wins? Founders cashing checks at unicorn vals. VCs? Sidelined, fees drying. Retail? Still locked out.
Look, AI’s reshaping everything. Chips like Positron could dent Nvidia’s grip—deployed at Oracle, not just hype. But betting the farm pre-revenue? That’s not strategy. That’s Valley fever.
We’ve been here. Dot-com 2.0, but with more GPUs. I’ll watch from the sidelines, sipping coffee, as the all-ins play out.
🧬 Related Insights
- Read more: Recursive Language Models: AI’s Recursion Fix for ‘Context Rot’ – Or Compute Nightmare?
- Read more: Railway’s $100M Gambit: Custom Data Centers to Supercharge AI Devs
Frequently Asked Questions
What is the AI gold rush doing to startup investing?
Family offices are bypassing VCs for direct stakes in private AI companies, chasing early wins as IPOs dwindle.
Are family offices better at picking AI winners than VCs?
Doubtful—they lack VC’s failure-forged radar, but bring deep pockets and patience for long holds.
Will direct AI bets pay off big?
Maybe for a few. Most? Echoes of dot-com busts, with sky-high risks in frothy private markets.
