🍄 DAILY SHROOMS

ServiceNow Drops $7.75B on Armis - AI Security Gets Serious

ServiceNow just made its biggest acquisition EVER, scooping up cybersecurity startup Armis for a cool $7.75 billion yesterday. Why's this a big deal? Because as AI systems get baked into everything from your coffee maker to hospital networks, the security stakes are getting higher than Snoop Dogg at a Willie Nelson concert. Armis specializes in detecting vulnerable devices across networks - something absolutely critical as we've got billions of AI-powered gadgets connecting to corporate systems. ServiceNow CEO Bill McDermott is essentially saying: 'Look, we're already the platform running your business workflows, now we'll also make sure your fancy new AI tools don't become a backdoor for hackers.' The timing couldn't be better – just yesterday, Uzbekistan's entire license plate surveillance system was found completely exposed online without even basic password protection. Oops! The bottom line: As AI systems get more powerful, securing them becomes exponentially more important. ServiceNow just paid a premium to own that problem.

• POW!Alphabet just dropped $4.75B to buy Intersect, showing Google's parent company is dead serious about securing the electricity it needs to power its increasingly power-hungry AI systems.
• ZOOM!Researchers released 'SAGE' this week (arXiv:2512.17102) - an RL framework that lets AI agents build and reuse skill libraries, significantly boosting task completion rates while cutting interaction steps.
• BAM!The US is holding off on new Chinese chip tariffs until at least mid-2027 in what analysts are calling a strategic pause in the ongoing AI semiconductor cold war.
• WHOA!France's postal service La Poste confirmed yesterday it's been hit with a suspected DDoS attack that's disrupted both mail and banking operations across the country.

Armorly

Armorly Just released in the Chrome Web Store, this free open-source extension blocks AI-native advertising embedded directly into chatbot responses - ads that traditional blockers like uBlock Origin can't catch. It intercepts SDK initialization from emerging ad networks (Koah, Monetzly, Imprezia) and strips affiliate tracking from AI-generated links. With AI companies exploring monetization through in-response ads, this is worth adding to your privacy toolkit before those "recommendations" start coming with sponsored fine print.

FiberCop

KKR-backed FiberCop is throwing down the gauntlet against Italy's broadband agency, filing a legal challenge yesterday alleging the agency gave preferential treatment to FiberCop's main competitor in the country's €4 billion fiber network deployment. While not a traditional 'success' story, it highlights how AI's insatiable appetite for bandwidth is making fiber infrastructure the new oil. Companies are fighting for control of the digital highways that will power tomorrow's AI applications, and FiberCop isn't willing to be sidelined without a fight. The outcome could reshape Italy's entire AI readiness landscape.

Solomonoff-Inspired LLMs: Making AI Less Confidently Wrong

A breakthrough paper dropped on arXiv yesterday (ID: 2512.17145) that could make your AI less confidently wrong. Researchers introduced a Solomonoff-inspired method for reasoning under uncertainty that distributes probability more evenly across competing hypotheses. In plain English: unlike current models that often double down on their first guess, this approach makes AI more conservative and aware of what it doesn't know. For developers building critical apps, this translates to fewer hallucinations and better performance on uncertain tasks. The method shows particular promise for financial forecasting, medical diagnosis, and legal reasoning - anywhere being wrong with high confidence is a disaster. If you're building systems that need to know when they don't know, this paper is worth implementing in your production stack today.

AI's Power Problem Just Got Real - Literally

Alphabet's $4.75B acquisition of Intersect yesterday isn't just another tech deal - it's a warning sign about AI's biggest bottleneck: electricity. Google's parent company is spending billions not on chips or talent, but on the actual power generation needed to run its AI systems. This mirrors Microsoft's recent long-term power purchase agreements and Amazon's massive investments in renewable energy infrastructure. The smart money is saying loud and clear: the AI companies that control their own energy destiny will win the next phase of the AI race. With AI models doubling in size roughly every 6 months, yesterday's deal suggests Google expects to need at least 2-3x its current power capacity by 2027. For investors, energy infrastructure might be the sleeper AI play of 2026.