🤖 Google to Invest Up to $40 Billion in Anthropic

🤝 Google to Invest Up to $40 Billion in Anthropic

🐳 China’s AI Whale Resurfaces With BIG Breakthrough

📍 6 AI Stock Sectors

🛠️ 5 Trending Tools

🥪 4 Brief Bites

💰 3 Funding Frontlines

💼 4 Job Opportunities

This strategic investment involves an initial $10 billion in cash, with the remaining $30 billion contingent on certain unspecified commercial performance milestones. Google also plans to provide additional “compute support” alongside this funding arrangement, supplying 5 GW of additional compute capacity over the next five years.

It’s all part of a long-running relationship, with Google having already invested just over $3 billion in Anthropic since April 2023. For context, Anthropic already taps into millions of TPUs: purpose-built AI accelerators engineered to supercharge AI inference at scale. Imagine asking Claude: “Summarize this article in three bullet points!” AI inference is the logic and reasoning Claude performs to generate that bulleted summary.

Amazon and Anthropic also expanded their strategic collaboration, with the online retail giant committing up to $20 billion and providing up to 5 GW of compute capacity in exchange for Anthropic’s pledge to spend more than $100 billion over the next ten years on AWS.

Anthropic recently raised a $30 billion Series G funding round at a $380 billion post-money valuation. At the time, 500 business customers were spending roughly $1,000,000 annually on Claude. Today, that number reached 1,000 business customers, doubling in less than two months. The additional compute capacity will help support Anthropic’s growing Enterprise AI adoption.

These strategic collaborations are being nicknamed “AI’s infinite money glitch,” where Big Tech companies and frontier AI firms effectively inflate their valuations by circulating money among themselves. So, is this normal? Yes! It’s a form of the “multiplier effect”: a dollar isn’t spent just once. The concern here is whether the circulating money is backed by real demand.

A context window refers to the number of tokens a frontier AI model like OpenAI’s “GPT-5.5,” which powers ChatGPT, can process or remember at once when generating a response. A token is a word or part of a word. For example, “yesterday” might be broken into three tokens: “yes,” “ter,” and “day.” In English, 1,000 tokens is roughly 750 words.

Typically, frontier AI models leverage attention mechanisms to calculate how much each word should pay attention to every other word across a conversation to capture the context. In other words, how much each new token should pay attention to every other existing token within the context window to determine what’s important when generating a response.

In “DeepSeek-V4 Preview,” a more complex algorithmic architecture called CSA takes a group of four tokens and squashes them into one compressed KV Cache, which enables it to retain, recall, and reuse information from earlier parts of the context window to reduce redundant calculations when generating a response.

Imagine reading a lengthy book and trying to remember every word on every page. Instead, CSA focuses on the most important groups of words to grasp the lengthy book’s plot as efficiently as possible.

With the typical attention mechanisms, a 100,000-word document costs 100x more to process than a 10,000-word document. With DeepSeek’s “compressed tokens,” a 100,000-word document only costs 10x more to process.

Era raised a $9M Seed Round for AI-driven object intelligence.

Sereact closed a $110M Series B for AI-powered warehouse robotics.

Ineffable Intelligence landed a $1.1B Seed Round for AI superlearners.

Tesla (Palo Alto, CA): AI Engineering Intern, Data Engine, Summer 2026

NVIDIA (Seattle, WA): Software Engineer, AI Networking, Entry-Level

xAI (Memphis, TN): Data Analyst, Physical Infrastructure, Mid-Level

OpenAI (London, UK): Partner AI Delivery Engineer, AWS, Senior-Level