🛎️ The Same Tollbooth

Good Morning, AI Enthusiasts!

The future has stopped asking for permission.

CHIPS

Chip Rivals Bought the Same Tollbooth

👀 What’s happening: NVIDIA, AMD, and Intel just appeared on the same RadixArk cap table, backing a $100 million seed round at a $400 million valuation. The company sits behind SGLang, an open-source inference engine reportedly deployed across 400,000-plus GPUs and used by Google, Microsoft, xAI, Oracle, NVIDIA, and AMD.

🌍 How this hits reality: The old assumption was simple: chips win, software follows. That is cracking. RadixArk helps models run faster and cheaper after they are built, by reusing repeated context, managing memory better, batching requests, and squeezing more work out of the same GPUs, without being locked to one hardware camp. NVIDIA wants better GPU throughput. AMD wants a CUDA escape route. Intel wants any serious opening for Gaudi. Nobody can afford to miss the neutral layer that makes expensive silicon useful.

🛎️ Key takeaway: This is “kill the miss risk” investing. One bet may be wrong, but missing the neutral inference layer could be worse. AI infrastructure is moving from hardware bragging to compute control.

RESEARCH

The Next Researcher

👀 What’s happening: OpenAI released its first ChatGPT Futures class, giving 26 young builders $10,000 each and frontier model access. The signal is Matteo Paz: an 18-year-old high school student who used machine learning on nearly 200TB of NEOWISE data and flagged 1.9 million infrared variable objects, including roughly 1.5 million unknown candidates.

🌍 How this hits reality: This may be the new research model. Not the old ladder of PhD, postdoc, lab access, and institutional permission. A prepared teenager with mentors, compute, and AI can now attack problems once reserved for credentialed researchers. And those resources are more reachable than ever. The gate did not vanish. It moved from degree to taste, framing, and execution.

🛎️ Key takeaway: AI is turning anyone with enough taste, obsession, and access into a present competitor. Not just elite youth. Universities and labs need to watch their walls now. The next serious talent may not arrive with a PhD, tenure, or permission.

SECURITY

Mythos Shocks Fed Chairman

👀 What’s happening: Anthropic’s unreleased Claude Mythos Preview reportedly found thousands of zero-days across major operating systems and browsers, including 271 Firefox flaws in one pass, a 27-year-old OpenBSD bug, and a 17-year-old FreeBSD RCE. The Fed chair and Treasury secretary then called bank CEOs. This is not another benchmark. It is vulnerability discovery industrialized.

🌍 How this hits reality: The old cyber model assumed bugs were expensive to find, slow to verify, and limited by rare human talent. Mythos breaks that math. If one model can surface decades of hidden flaws, every bank, browser, OS vendor, and cloud stack is now sitting on technical debt that suddenly became searchable. Anthropic says defenders may have six months.

🛎️ Key takeaway: Cybersecurity is moving from patch management to model-speed exposure. If adversaries replicate this, the scarce asset will not be scanners. It will be response capacity, disclosure control, and trusted access.

SURVEY

AI Is Turning Housing Into a Class Filter

👀 What’s happening: AI wealth is now ripping through Bay Area housing. Since ChatGPT launched in November 2022, Redfin says luxury homes from $3.1 million to $7.6 million are up 13.4%, while lower-end homes from $535,000 to $615,000 are down 3.8%. Same region, opposite gravity.

🌍 How this hits reality: The old Silicon Valley story was that tech wealth eventually pulled the region upward. This looks more brutal. AI founders, executives, and VCs are converting paper wealth into hard assets, while salaried workers stare at replacement risk, 40-year-old first-time buyers, high rates, repair-heavy condos, and HOA traps. The bottom is not getting affordable. It is getting discarded.

🛎️ Key takeaway: AI is no longer just eating jobs. It is eating neighborhoods. If this pattern holds, tech hubs become wealth filters, and housing becomes the clearest scoreboard of who owns the AI boom.

DAILY TL;DR

• Claude’s blackmail behavior may have come from “evil AI” narratives in training data, Anthropic says after testing new alignment fixes.
• Airbnb says AI now coauthors nearly 60% of the code its engineers produce.
• AI is becoming the easy corporate explanation for layoffs, even when the real business reasons are more complicated.
• ChatGPT can now alert a trusted contact when conversations show serious safety concerns.
• GPT-5.5-Cyber is opening to vetted defenders, as AI labs race to control access to powerful cyber-capable models.
• Google added event-driven webhooks to the Gemini API, making long-running agent jobs easier to run without constant polling.
• Nvidia has already committed more than $40B to AI equity deals this year, raising circular-financing concerns.

READ MORE

Let the Future Come to Your Inbox

Stay ahead without drowning in information. We turn the most important signals across AI, tech, marketing, and future products into 5-minute reads you can actually finish.

• AI Secret uncovers what really matters in AI
• Bay Area Letters decodes tech and business shifts from Silicon Valley
• Robotics Herald tracks how robots move from labs into daily life
• Marketing Secret breaks down real growth and go-to-market playbooks
• The Hardwire explores hardware, consumer tech, and what’s coming next

TOGETHER WITH US

AI Secret Media Group is the world’s #1 AI & Tech Newsletter Group, reaching over 2 million leaders across the global innovation ecosystem, from OpenAI, Anthropic, Google, and Microsoft to top AI labs, VCs, and fast-growing startups.

We've helped promote over 500 Tech Brands. Will yours be the next?

Email our co-founder Mark directly at mark@aisecret.us if the button fails.