Google paid $2.4 billion to license Windsurf's code and hire its founders. Four months later, they launched Antigravity, which developers instantly recognized as rebranded Windsurf. Here's the bizarre story of how one startup got split between three tech giants.

Forty-seven seconds.

That's how long it took Abhinav Gupta to spot Google's $2.4 billion secret.

He downloaded Antigravity on November 18th. Google's "revolutionary" new AI coding IDE. The one they'd been hyping at the Gemini 3 event. He opened it. Started typing. Then stopped.

Wait.

He'd used this interface before. The autocomplete felt familiar. The agent behavior is identical. Even the settings menu looked the same. Four months earlier, it was called Windsurf.

"Google just repackaged Windsurf as Antigravity," he posted on LinkedIn.​

Within hours, developers were dissecting screenshots. Comparing UI elements pixel by pixel. Testing features side by side. The evidence mounted fast. Same architecture. Same workflows. Same internal naming conventions.​

Google had spent $2.4 billion on this "new" product. But it wasn't new at all.

Two Friends, One Dream

Varun Mohan and Douglas Chen met in middle school.​

By MIT, they were inseparable. After graduation, their paths diverged — Chen at Meta building Oculus Quest VR, Mohan diving into GPU optimization. But they stayed connected. The startup itch never went away.​

2020: They founded Windsurf. Not as an IDE initially. As a GPU optimization tool. But by 2024, they'd pivoted hard into AI-powered coding.

The timing was perfect.

Developers were desperate for alternatives to expensive tools like Cursor. Windsurf offered something rare: a free tier that actually worked.​

By early 2025, they'd built something real. Two hundred fifty employees. Eighty-two million dollars in annual recurring revenue. Three hundred fifty enterprise customers. Not projections. Not promises. Actual revenue. Actual customers.​

This is the dream, right? Middle school friends building an $82M company together. But in Silicon Valley, success attracts attention. And not always the good kind.

When Big Tech Comes Calling

April 2025. OpenAI made the first move.

Three billion dollars to acquire Windsurf outright. Mohan and Chen entered exclusivity. No talking to other suitors. Just OpenAI, lawyers, and term sheets. The deal looked solid.​

June arrived. The exclusivity period expired.

The deal fell apart.​

No official explanation. No press release. Just silence.

Mohan and Chen went back to building. But the word was out: Windsurf was in play. And when a $3 billion deal collapses, everyone wants to know why.

Google saw an opportunity.

Friday, July 11, 2025: The Unraveling

Eleven AM Pacific. The announcement dropped.​

"Google hires Windsurf CEO Varun Mohan, co-founder Douglas Chen, and select R&D team in $2.4 billion deal."

Read that again. Not "acquires Windsurf." Not "buys the company."

Hires the team. Licenses the technology. Non-exclusive.​

The 250 remaining Windsurf employees learned about it like everyone else. On Twitter. In Slack channels. Through panicked texts from friends asking "What happens to you now?"

Nobody knew.

Mohan gone. Chen gone. The core R&D team gone. But the company still existed. The product still ran. Customers still paid.

Imagine being one of those 250 people. Your CEO just left for $2.4 billion. Your company is in chaos. And you're reading about it on Twitter.

Five PM. Russell Kaplan from Cognition AI picked up the phone.​

He called Windsurf's remaining executives. "We want to acquire what's left."

The Weekend That Changed Everything

Friday night through Sunday, the Windsurf team negotiated.​

Multiple suitors circled. Everyone wanted a piece. The IP. The brand. The customer list. The 250 employees wondering if they'd have jobs Monday morning.

Cognition moved fast. They'd just launched Devin, an AI coding agent generating massive hype. Acquiring Windsurf's technology and team would supercharge their roadmap. Plus, they needed to act. Before someone else did.​

The math was brutal but clear.

Google had paid $2.4 billion for talent and a license. They didn't own the product. They didn't own the trademark. They didn't own the customer relationships.​

All of that was still up for grabs.

Monday morning, July 14: Cognition signed the deal. Every remaining employee. Every line of code Google hadn't licensed. Every customer contract. The Windsurf brand itself.​

Jeff Wang became interim CEO. His first message to the team: "This has been a really volatile time".​

Understatement of the year.

Cognition promised respect. They waived vesting cliffs. Everyone got paid. Nobody got fired.​

But here's what nobody understood yet: Google's "non-exclusive license" meant Windsurf could keep operating. The original product would survive. It would compete.​

Against Google's version. Built from its own codebase.

Three companies now owned pieces of one startup. Google had the founders and the code license. Cognition had the product and the team. OpenAI had nothing but a failed $3 billion bid.

And the strangest competitive dynamic in tech history was about to begin.

Four Months of Silence

Google went dark. No updates. No previews. No leaked screenshots.

Meanwhile, Windsurf under Cognition kept shipping. Bug fixes. Feature updates. New AI model integrations. They acted like nothing had happened.​

Because from their perspective, nothing had, except new ownership and a competitor building with their old playbook.

Developers wondered: What was Google doing with that $2.4 billion license?

November 17, 2025: The answer arrived.

The Reveal

Google's Gemini 3 event. Big stage. Massive audience. Sergey Brin himself showed up to demo features.​

"Introducing Antigravity. Our agent-first AI coding platform".​

The demo looked slick. Autonomous agents. Multi-file editing. Browser-based testing. Gemini 3 Pro integration. It topped the LMArena leaderboard with a 1,501 Elo score.​

Developers downloaded it immediately.

Opened it.

Started typing.

And within minutes, Reddit exploded.​

"This is literally Windsurf."

"Same UI. Same keybindings. Same agent workflow."

"Google didn't build this. They rebranded it."

Screenshots flooded Twitter. Side-by-side comparisons. UI elements pixel-for-pixel identical. Settings menus with the same options in the same order. Even error messages using identical phrasing.​

One developer posted a forensic analysis: "The .antigravity config files have the same structure as .windsurf files. They just did a find-and-replace".​

Here's where the story gets darkly funny.

Google paid $2.4 billion. Hired world-class engineers. Had four months to build something revolutionary.

And they shipped Windsurf with Gemini 3 slapped on top.​

The comparison sites confirmed it. Both tools: 91/100. Nearly identical scores across every category.​

Because they were nearly identical products.

The Uncomfortable Competition

Here's the twist: Windsurf didn't die.

November 24 , seven days ago , Jeff Wang and the Cognition team shipped a major update. Claude Opus 4.5 integration. GPT-5.1 support. New features. Active development.​

The original product. Competing against Google's licensed clone.

Both using the same core technology. Both targeting the same developers. Both claiming to be the best AI IDE.​

And then there's Cursor.

The $20/month incumbent watching this circus unfold. Their community grew fiercer. "I'd rather pay than deal with this drama," one user wrote on Reddit. "Cursor is worth every penny for the peace of mind".​

Developers now face the weirdest choice in coding history:

Windsurf ($15/month): The original team minus their founders. Built by the people who survived the acquisition. Scrappy. Independent. Still shipping weekly.​

Google Antigravity (free for now): The licensed clone. Backed by infinite Google resources. Gemini 3 integration. Enterprise support. But also: buggy, experimental, and built from someone else's playbook.​

Cursor ($20/month): The Switzerland option. Completely separate. No acquisition drama. Loyal community. Proven track record. The most expensive.​

All three use Claude 3.5 Sonnet and GPT-4o under the hood. Performance differences are minimal. The technology is nearly identical.​

So the choice comes down to philosophy. Not features.

Who do you trust? The original team? The tech giant that "borrowed" their work? Or the competitor who stayed out of the mess?

What This Really Means

This isn't just an IDE story.

It's a blueprint.

Google just showed every tech giant how to "acquire" a startup without triggering antitrust reviews. Don't buy the company, that's expensive and legally messy. Just hire the founders. License the technology non-exclusively. Build your own version. Slap your brand on it.​

It's cheaper. It's cleaner. And apparently, it's perfectly legal.

Varun Mohan now works at Google DeepMind. He's helping them compete against Windsurf — the company he founded, using code he wrote, with a brand he built. His former employees maintain that product under different ownership. His old customers pay Cognition instead of him.​

Everyone got paid. Nobody went to jail. The startup didn't exactly die.

It just got dissected. Like a biology lab frog. With three companies each taking an organ.

Here's what keeps me up at night: If the team that built something can be hired away, leaving everyone else to maintain it while fighting a clone who actually owns the innovation?

Is it the people who wrote the code? The company that holds the license? The team that maintains it now?

I don't have an answer. I'm not sure anyone does.

The Unresolved Ending

Six months from now, one of three things happens:

Scenario 1: Google Antigravity adds enough Google Cloud integration that it becomes the obvious choice for enterprises. Windsurf gets squeezed out. Cursor survives on indie developer loyalty.

Scenario 2: Developers realize Antigravity is vaporware with a Google logo. Windsurf proves the original team knows the codebase better than Google's hired guns. Cursor keeps growing because nobody trusts the drama.​

Scenario 3: All three survive in different niches. Antigravity for Google Cloud shops. Windsurf for mid-market teams. Cursor for developers who'll pay $20/month to avoid thinking about acquisition politics.

I don't know which future we're headed toward.

But I know this: Developers on Reddit are sharing screenshots. Developers on Twitter are posting comparisons. Developers on Hacker News are arguing about who deserves their money.​

And somewhere in Google DeepMind, Varun Mohan is watching his old product compete against his new employer's version of that same product.​

That's either the ultimate success story or the strangest betrayal in tech history.

Maybe both.

The $2.4 Billion Question

When you license a codebase but don't buy the company… when you hire the founders but leave the team… when you build a "new" product that looks exactly like the old one…​

What have you actually acquired?

Google's bet: The future.

Windsurf's bet: A copy.

Cursor's bet: A mess.

In 72 hours, three companies carved up one startup. Four months later, we're still figuring out who got the best deal. And developers are stuck choosing between three nearly-identical products with wildly different backstories.​

Welcome to the AI era.

Where the code matters less than who owns the team that wrote it. Where "building something new" sometimes means taking something old, changing the name, and calling it innovation.​

The acquisition model is dead. The "acqui-license-and-clone" model is here.​

And I can't tell if that's brilliant or terrifying.