The $44 Billion Tweet That Broke Everything: How Elon Musk Bought Twitter on a Joke — Then Fired 80% of the Company in a Bathroom

The Text Message That Should Have Been a Joke

It was 2:48 AM on April 14, 2022. Elon Musk, sitting in his Austin compound, sent a text message to Twitter's board chairman Bret Taylor: "I made an offer."

Not "I'm considering an offer." Not "Let's talk about a deal." Past tense. Done.

The offer was $54.20 per share — a number Musk chose because it included "420," a marijuana reference he thought was funny. The total price: $44 billion. For a company that had never turned a meaningful profit. For a platform Musk had spent years trolling, meming, and nearly getting banned from.

Twelve hours later, Twitter's board activated a "poison pill" to block him. Six days after that, they accepted his offer. And for the next six months, Musk would try desperately to escape the deal he'd made on a whim.

But this isn't a story about a billionaire's buyer's remorse. This is a story about what happens when someone treats a 16-year-old company with 7,500 employees like a failing startup — and discovers that some of it actually was failing.

The Company That Couldn't Make Money Selling Ads

Twitter in 2022 was a paradox. It had 237 million daily active users. It drove global news cycles. World leaders used it to start wars and end them. It had become the de facto public square of the internet.

It was also hemorrhaging money.

From 2019 to 2021, Twitter lost a combined $1.1 billion. Its ad platform was so broken that advertisers preferred buying ads on Reddit — a site with half the traffic and a userbase famous for hating ads. Twitter's infrastructure ran on over 1,200 microservices that engineers called "the Rube Goldberg machine." Deploying a simple feature could take weeks because nobody understood how the services talked to each other anymore.

Former CEO Jack Dorsey had been effectively part-time since 2015, splitting his attention between Twitter and Square (now Block). His replacement, Parag Agrawal, had been CEO for exactly four months when Musk texted him: "Is Twitter dying?"

Agrawal, a Stanford PhD who'd risen through Twitter's engineering ranks, responded with a thread about daily active user growth and engagement metrics. Musk replied: "What did you get done this week?"

That exchange, leaked later, captured everything. Agrawal saw Twitter as a complex sociotechnical system that needed careful optimization. Musk saw a bloated company that needed a flamethrower.

The Bathroom Summit

On October 26, 2022, Elon Musk walked into Twitter's San Francisco headquarters carrying a sink.

"Let that sink in," he tweeted, posting a video of himself grinning in the lobby. It was the kind of dad joke that would make a middle manager cringe. But Musk wasn't joking about what came next.

By 5 PM, CEO Parag Agrawal was fired. So was CFO Ned Segal, CLO Vijaya Gadde, and General Counsel Sean Edgett. All four were escorted out by security. Their combined severance packages would eventually cost over $200 million.

Musk's first all-hands meeting happened two hours later — in a bathroom. No, really. He gathered about 30 engineers in Twitter's seventh-floor bathroom (it was one of the few rooms without windows or cameras) and asked them to explain what they did.

One engineer, a backend specialist who'd worked on Twitter's recommendation algorithm for three years, began explaining the machine learning pipeline. Musk interrupted: "Why does it take so long to load the timeline?"

The engineer explained the complexity: 1,200+ microservices, multiple data centers, machine learning models running inference on billions of tweets, compliance requirements, abuse prevention—

Musk cut him off again. "What if we just deleted half of it?"

The room went silent. You can't just delete half of a production system. Right?

The Code Review at Midnight

That weekend, Musk sent an email to all engineering staff: Print out your last 30 days of code commits. Bring them to a conference room on Monday. Be prepared to explain what each function does.

What followed was a 48-hour code review that felt more like a trial. Engineers sat across from Musk and his two-person "transition team" — investor Jason Calacanis and entrepreneur David Sacks, neither of whom had worked at Twitter or understood its codebase.

Musk would point at a function and ask: "What does this do?"

If the engineer couldn't explain it in one sentence, Musk would say: "Delete it."

If they explained it but it sounded like optimization for an edge case, Musk would say: "Delete it."

If they explained it and it touched compliance, content moderation, or accessibility: "Delete it."

One senior engineer, who'd worked on Twitter's abuse prevention systems since 2016, tried to explain why you can't just remove rate limiting without opening the platform to spam bots. Musk's response: "Then the bots will expose themselves faster. Delete it."

By Wednesday, November 2nd, Twitter had sent layoff notices to 3,700 employees — roughly 50% of the company. But that wasn't the real number. Over the next two weeks, another 2,000 would quit, fired, or let go. By December, Twitter had lost 80% of its workforce.

The engineering team that had maintained 1,200 microservices went from 1,500 people to under 200.

The Microservices Purge

Here's where the story gets weird: Twitter didn't break.

Elon Musk and his skeleton crew began systematically shutting down microservices. Not refactoring them. Not migrating them. Just... turning them off.

They started with the lowest-traffic services. The ones handling features nobody used. The ones created by long-gone teams for A/B tests that ended in 2017. Twitter's codebase was like an archaeological dig — layers of infrastructure built for problems that no longer existed.

One service handled serving different versions of the timeline to users in South Korea based on their telecom provider. It was built in 2015 for a partnership that ended in 2016. It ran on six servers and consumed 2% of Twitter's compute budget. Musk's team turned it off on a Friday night.

Nothing broke.

They turned off 50 more services that weekend. On Monday, site reliability engineers checked the dashboards. Latency was actually down. Turns out, all those microservices calling each other added overhead.

By January 2023, Twitter had shut down over 400 microservices. The infrastructure cost dropped from $1.3 billion a year to under $500 million.

But here's what nobody talks about: Twitter also lost features. The site became visibly less polished. Bugs that would've been caught in code review shipped to production. Accessibility features broke. Content moderation slowed to a crawl. The Android app started crashing on specific device models because the team that maintained device compatibility was gone.

Musk's bet was that users wouldn't care. And for the most part, they didn't. Twitter's daily active users actually grew during the chaos — partly because the chaos made Twitter the story.

The Turning Point: When the Whale Came Back

On December 28, 2022, at 3:47 PM PST, Twitter went down.

Not partially down. Not slow. Down. The app wouldn't load. The website returned a 503 error. And on screens around the world, users saw something they hadn't seen in years: the Fail Whale, Twitter's infamous error mascot from the early 2010s.

Except it wasn't the Fail Whale anymore — it was just a generic error page. But the symbolism was obvious. Twitter, the platform that had survived the Arab Spring and the Trump presidency and countless viral moments, had been taken down by... a skeleton crew rebooting a database cluster.

The outage lasted 73 minutes. When Twitter came back online, Musk tweeted: "Works on my phone."

Inside Twitter's half-empty offices, the remaining engineers were panicking. They'd been running on adrenaline and fear for two months. The outage was a warning sign: they were this close to a catastrophic failure.

Elon Musk's response? He hired back over 100 engineers who'd been laid off or quit — but only on 3-month contractor terms. No benefits. No equity. No job security. Just cash and the knowledge that if they didn't stabilize the site, their contractor checks would stop.

It worked. Barely.

The Legacy: The Billion-Dollar Efficiency Experiment

By mid-2023, Twitter had stabilized. The site ran on a fraction of its previous infrastructure. Costs were down 60%. The remaining engineering team had rebuilt critical services from scratch, consolidating 1,200 microservices into about 200.

But Twitter also lost something: its identity as a place that cared about quality, moderation, and accessibility. Bugs piled up. Misinformation spread faster. Features users loved — like the chronological timeline — became harder to access. Advertisers fled, spooked by Musk's erratic behavior and the gutted content moderation team.

Revenue dropped from $5.1 billion in 2021 to an estimated $3 billion in 2023. The company Musk bought for $44 billion was now worth, according to internal valuations, about $19 billion.

But here's the uncomfortable truth Silicon Valley won't admit: Musk proved something. Twitter was bloated. It did have too many microservices. It didn't need 7,500 employees to run a text-based platform.

Every tech company watched what happened at Twitter — now rebranded as X — and quietly asked their CFOs: "Could we do that here?"

Meta laid off 11,000 employees in 2022, then another 10,000 in 2023. Google cut 12,000. Amazon axed 27,000. Microsoft, Salesforce, and dozens of startups followed. The era of "growth at all costs" was over. Musk hadn't just bought Twitter. He'd shown an entire industry that you could cut half your workforce, delete half your infrastructure, and the site would keep running.

Whether that's innovation or vandalism depends on who you ask.

The Question That Haunts Silicon Valley

On a Tuesday night in March 2023, a former Twitter engineer — one of the early layoffs — sat in a brewery in San Francisco's Mission District. He was asked by a friend who still worked at Google: "Was Twitter really that inefficient, or did Elon just get lucky?"

The engineer stared into his beer for a long moment. Then he said: "Both. We had 400 microservices that did nothing. We also had 50 engineers who knew how to keep the site up when things caught fire. Elon deleted the 400 and fired 45 of the 50. Twitter's still running. But nobody knows what happens when the next fire starts."

That's the legacy of the $44 billion tweet. Not that Musk saved Twitter. Not that he destroyed it. But that he asked a question nobody in tech wanted to answer: How much of what we build actually matters?

And every quarter since, as tech companies announce layoffs and cost cuts and "efficiency initiatives," they're still trying to figure out the answer.