The case against liquidity

“Olga asks great questions, and her podcast is hands down the best.”

Ivana Delevska, Founder & CIO of Spear

Ivana Delevska, Paulo Passoni and I last week in NYC

This Week’s Essay

One of the smartest wealth-creation machines in the world is BTG Pactual, Brazil's most profitable major bank. BTG is built on a rule that sounds almost punitive. If a partner wants to leave, or just take some money off the table, they can't sell their shares on the open market. They sell them back to the partnership at book value. And the shares get reassigned to the remaining partners, also at book value.

Ouch. Especially when the bank compounds shareholder equity at roughly 26% a year on a book value of about $12 billion.

The system was designed to make leaving expensive and staying obvious. Partners are required to keep their own wealth in the firm's funds. The whole thing works against the impulse to desert a sinking ship — which, by the way, is exactly what held the bank together in 2015, when its founder was arrested and BTG nearly collapsed.

But the ship doesn't have to be sinking. It could be sailing ahead at breakneck speed — and the model matters just as much. The same rule that keeps people from fleeing a crisis keeps them from being poached at the peak. In every tech business, top talent is the most valuable asset there is, and the hardest one to hold onto. Especially when you're winning and everyone wants a piece of your team.

Which brings me to SpaceX. This Friday, thousands of SpaceX employees are expected to become filthy rich. Engineers who spent years accepting below-market salaries in exchange for stock are about to see that bet pay off.

Which is wonderful news. For the engineers. For the company, it's the beginning of a very expensive problem.

The day that stock turns liquid, the math of staying changes. For years, the only way to touch that wealth was to keep showing up. The equity was golden handcuffs — the reason you stayed through the brutal hours and the launches that blew up on the pad. Now the handcuffs come off. The money is real, soon to be spendable, and sitting in a brokerage account. And the engineer who was mission-critical yesterday can wake up tomorrow, look at a number with six zeros, and ask a question they never had the luxury of asking before: why am I still doing this?

Some will leave to start their own thing. Some will leave to do nothing at all. And even the people who stay won't stay for the same reasons. The work that felt like a mission on Monday feels optional by Friday.

You might say that SpaceX is different. Many employees really believe they are helping humanity become a multi-planetary species. Whether that mission proves stronger than the temptation of liquidity will be fascinating to watch.

Every extraordinary company eventually runs into the same question: how do you keep exceptional people motivated after they no longer need the money? Or put differently: how do you manufacture hunger in people who no longer need anything?

None of this is new to venture. The effect of money on founders and key talent is one of the things great investors worry about most. It's why they pay close attention to secondaries. How much is the founder cashing out? How much are they keeping? What changes once the money hits their account?

It cuts both ways. You don't want a founder running on empty. They should be able to buy a house, put their kids in a good school, and stop worrying about rent. Desperate founders make desperate decisions. But take too much off the table and something shifts. The same money that removes the fear can remove the fire.

I find it fascinating how directly the interests collide. What's good for the person is bad for the company. The individual wants security. The company needs drive. The individual wants optionality. The company benefits from commitment.

In venture, we built machinery to manage it. Founder shares vest over years. Investors hold rights of first refusal and consent rights over how much gets sold. Secondaries are capped and negotiated round by round. Lock-ups delay liquidity. BTG designed its own version of a high cost of exit. And one way or another, Elon Musk, Dario Amodei, and every founder building a generational company will have to engineer their own.

Every company loves to say it hires missionaries, not mercenaries. Mercenaries leave when the check clears. Missionaries stay for the mission. The next generation of AI and space companies is about to run one of the largest real-world experiments in motivation ever conducted. And we're about to find out how many missionaries Silicon Valley actually has.

Community Picks

1. On speed becoming the moat

For most of the software era, the moat was technology. Build something technically superior, defend it with engineering talent, and compound. AI is changing that equation. Models are improving so quickly and becoming so widely available that technological advantages are getting arbitraged away faster than ever. What matters now is how quickly a company can ship, learn, and adapt.

2. On fear as the first wave of AI adoption

The first wave of enterprise AI spend is not driven by ROI. It is driven by fear. Fear of being late. Fear of looking stupid. Fear that a competitor figures it out first. Companies are spending because they feel they have to understand AI before it makes them obsolete. But fear is not a durable budget line. After fear comes rationalization. The next phase of AI adoption will be much more brutal: which tools actually save money, which workflows actually improve, and which vendors survive once CFOs stop paying for experimentation and start demanding measurable returns.

3. On creativity becoming the final bottleneck

AI can collapse the cost of production. It can make the commercial cheaper, the movie cheaper, the research cheaper, the workflow cheaper. But it does not automatically make the idea better. When production costs collapse, creativity becomes more valuable, not less. The scarce person is no longer the one who can execute the task. It is the one who knows what should exist in the first place.

4. On the most important metric of the AI era

Ten years ago, $200,000 of revenue per employee looked impressive. Today, it's almost irrelevant. The best AI-native companies are already pushing toward $1 million, $5 million, or even $10 million of revenue per employee. The number itself matters less than the direction. If revenue is growing while headcount stays flat, something powerful is happening inside the organization. Revenue per employee is becoming a proxy for leverage — whether a company is using AI to eliminate complexity and amplify talent, or simply adding people every time revenue grows. The most valuable companies may not be the ones with the largest teams. They may be the ones generating extraordinary output with surprisingly few people.

5. On Latin America's unexpected AI advantage

As hyperscalers race to build the next generation of AI infrastructure, the bottleneck is shifting from semiconductors to the physical world: electricity, land, cooling, permits, fiber. That changes the map of who benefits. For decades, Latin America sat on the sidelines of the big technology infrastructure cycles. This one may be different. Brazil, Chile, Argentina, Paraguay, Colombia, and Panama have something increasingly scarce — abundant energy, available land, water for cooling, and the ability to build at a fraction of the cost of developed markets. In some cases, the data center economics are dramatically more attractive than in the US or Europe.

Check out my favorite moments in this highlight reel:

What I’m Loving

Read — Elon Musk (Walter Isaacson)

The definitive Musk biography. Isaacson shadowed him for two years — attended his meetings, walked his factories, and interviewed his family, friends, coworkers, and adversaries to answer one question: are the demons that drive Musk also what it takes to drive innovation and progress? With SpaceX going public this week, it's the right moment to revisit how the most consequential — and most polarizing — operator of our era actually works.

Read — The SpaceX S-1

Buried in the filing: the Starlink subscriber and ARPU numbers, the 99% cost-per-kilo reduction that underpins the whole business, and the plan to deploy orbital AI compute satellites as early as 2028 — handling energy-intensive AI workloads like inference in space, at greater scale and efficiency than terrestrial data centers. The data center race might not stay on Earth. Read it yourself.

Listen — Marc Rowan on Private Markets, Software Repricing, and Capital Allocation (The a16z Show)

Paulo flagged this one on the episode. Marc Rowan — cofounder and CEO of Apollo, maybe the best in the world at architecting the cost of capital for massive projects — walks through how private capital is becoming the engine that finances the real economy: data centers, energy, robotics, the entire AI buildout. If the data center bottleneck in this issue caught your attention, this is the conversation on who actually pays for it, and how the right financial engineering makes hundred-billion-dollar projects pencil out.