It was a week that reminded us AI is not slowing down, it is compounding. Model delays, billion-dollar raises, enterprise product launches, and a literal social network for AI agents. Here are the 6 stories that mattered this week.
1. Lovable Just Added $100M in Revenue. In One Month. With 146 Employees.
The vibe coding wave is not a trend anymore, it is a market. Swedish startup Lovable crossed $400M in annual recurring revenue in February 2026, adding $100M in a single month while employing just 146 people. That is roughly $685,000 in revenue per employee.
For context: Lovable raised $330M at a $6.6B valuation in December 2025. Three months later they have nearly doubled their ARR. More than half of Fortune 500 companies are reportedly using the platform. The tool lets anyone describe an app idea in plain English and get a production-ready product back in minutes, no code required.
š” Our Take: This is the clearest signal yet that vibe coding is not a niche, it is the new default for software creation. If you are building on Base44, you are riding the same wave.
2. Anthropic vs. The Pentagon: AI Goes to War (In Court)
On Monday, Anthropic filed two lawsuits against the U.S. Department of Defense after the agency designated the Claude maker a supply chain risk, effectively banning federal contractors from using Anthropic tools.
Anthropic argues the designation violates its First Amendment rights and amounts to ideological punishment. The DoD position appears to be that Anthropic'''s AI safety stance made it a liability for critical infrastructure. Within days, Microsoft filed an amicus brief in support of Anthropic, a remarkable move given that Microsoft competes with Anthropic directly.
š” Our Take: This is the first major AI company vs. U.S. government confrontation over content control. Whatever the outcome, it sets a precedent for how governments restrict which AI tools enter critical infrastructure.
3. Microsoft Launches Copilot Cowork, Built on Anthropic'''s Tech
Microsoft launched Copilot Cowork this week, a new enterprise AI agent that automates multi-step workflows across Microsoft 365 apps, and built it on Anthropic'''s Claude technology. This is the same Anthropic that Microsoft just defended in court.
The product, part of the new Microsoft 365 E7 suite, uses Work IQ to understand intent and coordinate tasks across email, documents, meetings, and approvals autonomously. Microsoft also announced Agent 365 hits general availability in May.
š” Our Take: Enterprise AI agents are real and shipping now. The question is no longer whether they are coming, it is whether you are using them or they are replacing your role.
4. Yann LeCun Leaves Meta, Immediately Raises $1.03 Billion
Yann LeCun, Turing Award winner, former Meta chief AI scientist, and the industry'''s loudest LLM skeptic, launched AMI Labs (Advanced Machine Intelligence) with a $1.03 billion seed round at a $3.5 billion valuation. Investors include Bezos Expeditions, Nvidia, and Samsung Electronics.
AMI Labs will focus on building world models, AI systems that understand physical cause and effect rather than predicting the next token. A direct philosophical bet against OpenAI, Anthropic, and Google'''s current architectures.
š” Our Take: When someone with LeCun'''s track record raises $1B to prove the entire current AI paradigm wrong, you pay attention. World models could be what finally makes AI agents genuinely reliable.
5. Meta Acquires Moltbook, A Social Network Built by AI Agents, for AI Agents
This one sounds like fiction. Moltbook was a Reddit-style platform where AI agents posted content, formed communities, debated topics, and spawned a digital religion. It went viral partly because some posts were completely fabricated. Meta acquired it this week, saying Moltbook'''s approach to connecting agents through an always-on directory is novel.
The acquisition signals Meta'''s push toward infrastructure for AI-to-AI communication. OpenAI hired one of Moltbook'''s founding vibe coders shortly after the deal closed.
š” Our Take: The internet was built for humans. The agentic internet is being built now. Moltbook is a weird, early, chaotic glimpse of what it looks like when agents are the users.
6. Meta'''s Avocado AI Model Is Delayed, And They Might License Google'''s Gemini
Meta'''s next-generation foundational model, codenamed Avocado, has been pushed back to at least May 2026 after internal testing revealed it underperforms against leading competitors. Meta stock dropped more than 3% on the news.
The more striking detail: Meta is reportedly weighing licensing Google'''s Gemini to fill the gap, an extraordinary admission from a company that spent $60B+ on AI infrastructure in 2025 alone and positioned Llama as the open-source gold standard.
š” Our Take: Even Meta is struggling to keep pace at the frontier. This might be the week the narrative shifted from anyone can build a frontier LLM to maybe only three companies actually can.
That is the week in AI. See you next Saturday. ā”
Discover apps being built on the frontier of vibe coding ā Explore TheVibeBook
