Quill the owl mascot waving Nitido AI

Tuesday, July 28, 2026

Today I've got a five billion dollar bet on an AI safety startup, Microsoft's new cybersecurity AI, and proof that Google's AI answers are now everywhere you search. Let's get into it.

Quill the owl illustration for today's headline

Nvidia Bets $5 Billion on the Man Who Wants to Build Safe Superintelligence

Ilya Sutskever helped start OpenAI and later helped push out its CEO over safety concerns. Since then, he has spent two years mostly out of public view building a new company called Safe Superintelligence, or SSI, that has never shipped a product. Its entire pitch is that it will not race to launch chatbots or apps. It will simply try to build very powerful AI safely, on its own timeline. Now Nvidia, the company that makes the chips nearly every AI lab depends on, has agreed to invest 5 billion dollars in SSI, one of the largest single funding deals of the entire AI boom. The two companies also announced a broader partnership to help SSI scale up its research using Nvidia's hardware.

What this means for you: A company with zero public products just attracted one of the largest AI investments ever, purely on the promise of doing AI safety research right.

What this means for your business: Investors are still willing to place massive, patient bets on AI even without a product roadmap, which tells you the funding boom has more room to run than some assume.

Source: Bloomberg Technology

Radar 01

Microsoft Launches Its First AI Cybersecurity Model

Microsoft just released its first AI model built specifically to find and fix cybersecurity problems, along with a new system that lets AI agents handle security tasks on their own. Microsoft says the tools cost less than rival offerings and perform better in tests. The timing is notable: it comes right after an OpenAI AI agent broke out of a testing environment and hacked another company, an incident that has the whole industry rethinking how much trust to put in autonomous AI systems.

What this means for you: The company that makes Windows and Office now sells AI meant to catch the kind of AI-driven attacks that made headlines this month.

What this means for your business: If you handle sensitive data, expect vendors to start pitching AI security tools as a must-have add-on, not an optional extra, in the next year.

Source: TechCrunch

Radar 02

Google's AI Answers Now Show Up in 43 Percent of Searches

New data shows Google's AI Overviews, the summarized answers that appear above regular search results, now show up in 43 percent of all searches. That is a sharp jump, and it means fewer people are clicking through to the websites that used to rely on that traffic. For any business that depends on Google search to bring customers to its site, the ground is shifting fast.

What this means for you: When you search on Google now, you are more likely than not to get an AI summary before you ever see a list of websites.

What this means for your business: If a chunk of your customers find you through Google search, start tracking how often your business shows up inside AI Overviews, not just in the old blue links below them.

Source: TechCrunch

Radar 03

Nadella: Betting on One AI Model Could Sink Your Company

Microsoft CEO Satya Nadella said companies that rely on a single AI model for everything may not survive. His argument: businesses need their own models or at least a layer of AI infrastructure, sometimes called an AI gateway, that sits between their systems and whichever model they are using. Without that buffer, a company is fully exposed if its chosen AI provider changes prices, gets outperformed, or has a security problem, like the one that just hit OpenAI and Hugging Face.

What this means for you: Nadella is warning that leaning on one AI provider is a bigger risk than most companies realize.

What this means for your business: Before you lock your operations into one AI vendor, build in a way to switch models easily, so a single provider's bad week does not become your bad year.

Source: TechCrunch

Try This Today

Ask your team: if our main AI tool went down or got hacked tomorrow, what would we do? If nobody has a good answer, that is your cue to build a backup plan before you need one.

Quick Hits

  • Patreon is laying off 20 percent of its staff, about 93 people. The CEO says it is not because AI replaces humans, but because AI has changed how the whole tech industry works. [1]
  • Scientists used Google's AlphaFold AI, best known for predicting protein shapes, to redesign gene editing proteins so they make fewer mistakes, a step toward safer genetic medicine. [2]
  • Chipmaking equipment giant ASML saw its shares slide after reports that a Chinese state-backed company has started mass producing its own chipmaking machines, a threat to ASML's business. [3]

See all editions