Can Meta Buy Belief? - Michael Novati - Personal Essays
Michael Novati - Personal Essays
SubscribeSign in
Can Meta Buy Belief?<br>Facebook was built by people who acted like the product belonged to them. Meta’s AI future may depend on whether that kind of ownership can survive layoffs, forced transfers, and $100M engineers.
Michael Novati<br>May 31, 2026
31
Share
I’ve been thinking about Meta all week — not as an industry story, but almost as a grief story.<br>That sounds dramatic. Maybe it is. But Facebook wasn’t just a company I worked at. For a long time, it shaped how I thought about work, speed, honesty, standards, and what it means to actually care about what you build.<br>I still catch myself saying “we.”<br>So when I look at Meta now, I’m not just reading a strategy memo. I’m trying to decide whether I’m watching reinvention — or a company losing its way.<br>Meta has reportedly laid off thousands of people, moved thousands more into AI, and built a superintelligence lab filled with hires getting packages that, depending on the person, run into the tens or even hundreds of millions. Maybe that is exactly the right move. Frontier AI may require capital, compute, and a small number of world-class people operating at the edge.<br>But the risk isn’t the org chart.<br>The risk is the story everyone else tells themselves.<br>Are they part of the mission? Or are they the leftovers?<br>That’s the thing I keep coming back to. You can buy talent. You can buy compute. You can buy leadership.<br>Can you buy belief?<br>By belief, I don’t mean vibes. I don’t mean loyalty. I mean people doing the annoying, unassigned work because they feel like the product is theirs. The clean term is discretionary ownership. The lived version is simpler:<br>Nobody has to force you to care.<br>That was the engine of early Facebook.<br>I joined in 2009. There were ten of us in the new-hire class. Ten. The offices were scattered across downtown Palo Alto, and it didn’t feel like joining a giant company. It felt like joining a chaotic little organism that somehow already had hundreds of millions of users.<br>At my first Q&A with Mark, engineers were still frantically finishing the new Messages product minutes before showing it to the company. They were literally building the thing right up to the moment it was presented.<br>Then we moved to 1601 California Avenue, near Stanford, our first real campus. The place was ridiculous. RipStiks everywhere. People crashing into walls and dumpsters. Risk and chess games going until midnight, then turning back into code.<br>And underneath all the absurdity was the actual machine:<br>Two people handled Facebook Photos.<br>An intern ran Facebook Video.<br>A college kid was responsible for video at a company hundreds of millions of people used every day, and nobody thought that was weird, because everyone was carrying something they had no business carrying — and somehow carrying it anyway.<br>That leverage didn’t come from process. It didn’t come from a comp ladder. It came from people acting like nothing was someone else’s problem.<br>And it wasn’t just money. There was no public stock. I started at an $85,000 base. I lived in a hacker house with ten other people where the kitchen roof was caving in and we had bedbugs.<br>We worked insane hours, came home to that, and somehow life felt good.<br>Objectively disgusting. But good.<br>That’s the part that’s hard to explain if you only know the modern tech industry. People were there with their hearts.<br>I wrote a weekly internal blog in those years and became, as far as I could tell, the most-followed non-manager at the company. When I reread it now, plenty of it was naive and overwrought. But people read it because it said the corny thing out loud: the work mattered, and it was okay to care.<br>That doesn’t mean old Facebook was innocent. It wasn’t. It had arrogance, blind spots, and all the danger of young people moving too fast with too much power. Some of that danger created real consequences the company later had to answer for.<br>So I don’t want to romanticize it.<br>Belief is not virtue.<br>Belief is fuel.<br>The same ownership that built things overnight could also rationalize damage overnight. I’m not saying the old culture was morally pure. I’m saying it was an engine.<br>Now look at the words people use for Meta today: performance-obsessed, transactional, calibrated, hardened, elite.<br>Maybe that’s just what scale does. Maybe it’s what the AI fight requires. The brutal case for Meta is strong: frontier AI is not consumer social. A hacker house does not out-build OpenAI on nostalgia. Maybe Meta is assembling the best surgeons in the world and putting them in one operating room.<br>I concede all of that.<br>But a model is not a company.<br>A frontier lab can train something extraordinary. Turning it into products, infrastructure, trust, distribution, monetization, and policy is everyone else’s job. And most of those people are not getting nine-figure packages.<br>They are the ones who keep the systems running, ship the features, handle the...