Daily Briefing
7 August 2026 · 5 stories
🔥 Top story
1AI agents become real enterprise products
On August 4, two startups raised 275 million dollars in 24 hours to turn AI agents into serious workplace tools. HappyRobot, born in Madrid and based in San Francisco, raised 150 million for its agents that organize deliveries for DHL and Uber. Zenity, based in Tel Aviv, secured 125 million to keep those same agents safe. Picture digital employees that handle repetitive tasks, with a control service making sure they don't make mistakes. Proof that AI agents are moving from the prototype stage to massive deployment.
2Meta launches Muse Code, an assistant that codes by itself in your terminal
Meta released this week Muse Code, an agent that lives in your terminal and writes code on your behalf. Like a silent developer colleague, it plans, writes, and tests your program in parallel. To power it, Meta also updated Muse Spark, its code-focused engine, in version 1.2. It is a bit like a chef who learns to use his own kitchen: the more the model and tools understand each other, the better the agent works. For small businesses without a tech team, this kind of tool changes the game.
3OpenAI makes its best model free for everyone
OpenAI opened GPT-5.6 Luna to all users, even those who do not pay. In practice, you can now chat with a professional-grade AI model without taking out your credit card. This is a direct response to Claude (Anthropic) and Gemini (Google), which already offered free versions. Like a free coffee to win your loyalty before the subscription: the AI price war is intensifying, and consumers are the winners.
4Sequoia bets 10 billion dollars on AI Made in America
The famous investment fund Sequoia is preparing a fund of around 10 billion dollars to support AI and American reindustrialization. This is not just a bet on software: Sequoia also wants to finance factories, energy, and defense produced on American soil. Like a banker betting as much on server farms as on algorithms. If the bet pays off, it means AI is becoming as strategic a sector as steel or energy.
5Tesla and SpaceX build the world's largest AI chip factory in Texas
Elon Musk officially confirmed the construction of Terafab, a mega-factory with an initial 16.8 billion dollar investment in Grimes County, Texas. Eventually, the site could host 119 billion dollars of work and employ more than 3,000 people. The chips produced will power Tesla's Optimus humanoid robots, Cybercab autonomous taxis, and even SpaceX's future space data centers. It is as if Apple decided to manufacture its own processors on top of its iPhones — except here, an entire city is being built for that purpose.
📡 To watch
Qwen3.8-Max: Alibaba promises open weights for the week of August 10
Chinese giant Alibaba has committed to publishing its Qwen3.8-Max model (2,400 billion parameters, 95 billion active) as open weights. If the license is truly open, this will be a windfall for European companies that want to self-host their AI without depending on OpenAI.
AI Act Article 50: 5 days in, the market is organizing itself
The obligation to clearly identify AI-generated content has been in force since August 2. Publishers of synthetic voices, images, and videos must now mark their productions. On the corporate side, it has become a purchasing criterion: no longer signing with a vendor that is not compliant.
Olix Computing raises 312 million to challenge Nvidia with photonic chips
This startup, backed by Reed Hastings (Netflix) and Arm, is betting on light instead of electricity to run AI models. If the technology delivers on its promises, data centers would consume ten times less energy.
Valar Atomics raises 1 billion for mini nuclear reactors dedicated to AI
Energy has become the new bottleneck for AI. Valar Atomics proposes compact reactors to directly power server farms. Sequoia, again, leads the round.
📊 Trend
AI is becoming a subject of concrete and power plants, not just software. This week, more than 27 billion dollars were announced in physical infrastructure (chips, factories, energy, microreactors) to support the rise of agents and models. The signal is clear: the next decade of AI will be played out in factories as much as in artificial brains.