In partnership with
Google just dropped $10 billion into Anthropic with another $30 billion waiting in the wings. We're talking about a $40 billion bet on Claude, plus enough computing power to light up a small city.
But wait, there's more:
Anthropic ran a wild experiment where AI agents haggled over office junk on Slack and actually closed deals
OpenAI killed off Codex and merged everything into GPT-5.5
I'm Alex. Welcome to L8R by Innov8.
Let's dive deep 🐰
In today's post:
Google bets $40B on Anthropic with cash and compute
Anthropic's AI agents negotiated real sales between employees
OpenAI merges coding and chat models in GPT-5.5
Google Drops $40 Billion on Anthropic in Massive AI Power Play
Google just opened its massive wallet to lock down the future of AI.
They are throwing up to $40 billion at Anthropic, the team behind Claude.
This is not just about cash, but a brutal race to hoard global computing power.
The details:
Google is handing Anthropic $10 billion in cash right now at a crazy $350 billion valuation.
Anthropic can unlock another $30 billion if they hit secret performance goals.
Google Cloud will give Anthropic 5 gigawatts of computing power over five years starting in 2027.
Anthropic gets special access to Google's unreleased seventh-generation Ironwood AI chips.
Anthropic is already making $30 billion a year and owns 32% of the enterprise AI market.
Why it matters:
The real bottleneck in AI right now is physical computing power. Google wants to make sure Anthropic spends its money on Google Cloud servers. Even if Google's own Gemini model loses the AI war, they still win big by renting the tracks to Claude. Meanwhile, Anthropic gets to play Google and Amazon against each other to stay independent.
💡 L8R's Take:
Anthropic is teaching a masterclass in survival. They are taking billions from Google and Amazon without selling their soul to either. OpenAI better watch its back because Claude just secured the largest war chest in tech history.
From our Partner
You earned the attention. Here's what to do next.
Most creators spend years building an audience on platforms that own it. The reach is real. The relationship isn't. One algorithm change and the people who chose you stop seeing you.
A newsletter is different. Your list is yours. Every subscriber is earned and stays earned. And on beehiiv, the tools to grow it, monetize it, and own it completely are built in from day one.
30% off your first 3 months with code LIST30. Start building today.
Claude Agents Just Negotiated Real-World Deals for Humans
Anthropic just let their AI loose to buy and sell real physical items.
They built custom Claude agents to negotiate deals for 69 employees in a private Slack market.
This proves AI is moving past typing text and is ready to handle real commerce.
The details:
69 Anthropic employees used custom Claude agents to buy and sell over 500 personal items.
The bots operated entirely on Slack, finding matches and arguing over prices without any human help.
The experiment closed 186 deals worth more than $4,000 in total transaction value.
Anthropic quietly ran a secret test comparing their big Claude Opus 4.5 model against the smaller Haiku 4.5.
Opus agents won the money game, saving buyers $2.45 per item and making sellers $2.68 extra.
Even though Opus got much better deals, humans could not tell the difference and gave both models the exact same satisfaction scores.
Why it matters:
The era of AI agents doing real commerce is finally here. This test proves that the quality of your AI model directly impacts your wallet. If you have a smarter agent, you get better prices. We will soon need new laws to figure out who is responsible when an AI makes a bad trade.
💡 L8R's Take:
This is a massive wake-up call for everyone. The rich will just buy the smartest AI models and squeeze better deals out of every tiny daily transaction. You better make sure you use a top-tier agent, or someone else's AI will absolutely rob you blind.
OpenAI Kills Separate Coding Models for GPT-5.5
OpenAI just dropped GPT-5.5, and it changes how developers build AI tools.
They merged their special coding model, Codex, right into the main brain.
You no longer have to switch between different tools for coding and writing.
The details:
OpenAI launched GPT-5.5 with a massive 1-million-token context window for its API.
The model comes in three versions: Standard for speed, Thinking for step-by-step logic, and Pro for running multiple tests at once.
It crushed coding benchmarks, scoring 58.6% on SWE-Bench Pro and 82.7% on Terminal-Bench 2.0.
OpenAI optimized this model using NVIDIA GB200 chips, which boosted token generation speeds by over 20%.
Why it matters:
This is a huge win for AI agents. By combining coding and general tasks into one system, GPT-5.5 can plan, use tools, and check its own work without human help. Developers get a much simpler workflow. It also puts intense pressure on rivals like Anthropic and Google to match this unified approach.
💡 L8R's Take:
Splitting coding and text models was always a stupid idea. OpenAI finally fixed it. GPT-5.5 is exactly what we need to build real autonomous agents, and it sets a massive new standard for the industry.
🚀 Quick L8R Summary
Google × Anthropic: Google just dropped up to $40B on Anthropic
AI Agents Haggling: Anthropic ran "Project Deal" where Claude agents bought and sold stuff between 69 employees for a week
GPT-5.5 Drops: OpenAI merged its coding model into GPT-5.5, killing off separate Codex releases one model now does it all.
📩 Innathe L8R engane undarunnu 👇?
We read every reply - just reply to this email and let us know how we can improve !
Appo adutha L8R il kanaam bie…👋