Happy Sunday. Here's what actually mattered in tech this week, broken down so it's worth your five minutes, then a bit on what's been on my mind. Let's get into it.
This week in tech + AI
Google is quietly losing the people who built modern AI. Google DeepMind is going through a real shakeup. Jeff Dean, one of the most important engineers in the company's entire history, is reportedly leaving with a group of top researchers to start their own company, and there are reports that Demis Hassabis is stepping back from running DeepMind day to day. It's happening while OpenAI and Anthropic keep poaching Google's best people, and while the old split between Google Brain and DeepMind is finally being unwound. On paper Google has everything, the talent, the compute, the deepest research history in the field, and it still keeps watching its star people and its best product ideas walk out the door. What it means: having the best technology is not the same as winning. Even the company that literally invented the transformer can lose its edge if the people leave, and for anyone building a career in AI that's the real lesson, where the talent goes tells you more than any logo does. Read more
An AI model just did original math. OpenAI's unreleased model, Astra, solved ten open math problems that had been stuck for years, for around $2,000 in compute, and a separate OpenAI model disproved a famous 80 year old geometry conjecture from Erdős. The key part is these weren't textbook questions with known answers, this was genuinely new mathematics, and Fields Medalist Tim Gowers said he'd have recommended the proof to a top journal without hesitation. What it means: AI is crossing the line from answering questions to actually producing new knowledge. The skill that pays off now isn't memorizing facts, it's being able to frame the right hard problem and work with these tools to crack it. Read more
Oracle banned AI-generated code from one of the biggest projects in software. Oracle told contributors they can no longer submit AI-generated code to OpenJDK, the open source core of Java that quietly runs a huge chunk of the internet. What it means: even as everyone races to code with AI, the most serious codebases still want humans who actually understand what they're shipping. Good reminder that vibe coding has a ceiling, and knowing the fundamentals is still what gets you trusted with the important stuff. Read more
ChatGPT passed around a billion weekly users. OpenAI also opened unlimited text chats to free users and cut the price of its GPT-5.6 model by 80 percent. What it means: AI has basically become default infrastructure now, as normal as Google or email. When everyone has the same tools for free, the edge isn't having AI anymore, it's how well you actually use it. Read more
On my mind
What am I doing all of this for? That's a question I ask myself quite a lot, especially when I'm overwhelmed with work. And to be honest, I don't know the exact answer. I have no idea what my life will look like in a few years or what the exact goals are that I want to achieve. I just have this broad set of directions I want to follow: being financially free, having freedom over my time and where I live, and being technical enough to actually contribute to how society moves forward. Any of these can take many different forms, and I don't know yet how mine will look. But having these broad goals lets me keep moving even when I'm uncertain. Because I choose my habits and actions based on what pushes me in the general direction of where I want to be, without locking myself into one exact future that I might not even want anymore in a few months. That's why, for example, I'm confident that doing a technical masters, creating content and building a business are worth my time. Each one moves me toward that broad set of goals. I guess what you can take from this is that you don't need to have your exact goal figured out. A general direction of where things should be heading is often enough to keep you moving. Of course it's even better if you're super clear on your goals, but if you’re not, don’t worry :)
More from me
My latest video: How hedge funds use the math that was invented to build the atomic bomb. Watch here
Tool of the week: For everyone building with AI and wanting to keep track of AI spend, just clicking the Link supports me! :)
Know Exactly Who's Spending Your AI Budget.
Every AI request leaves a trail. Mesh gives engineering and finance complete visibility into who used which model, how many tokens were consumed and where your AI budget is going.
Stop guessing. Start governing AI spend.
Connect once, switch between GPT, Claude, Gemini and hundreds more whenever you want, while automatically routing requests for 40% lower costs and 99.99% AI response rate.
Have a great week,
Chris