Weekly Intel - 2026-01-11
The promises are starting to collide with reality.
This week brought a useful batch of evidence for anyone paying attention: cloud providers hiking prices with weekend announcements, hardware companies backing away from AI messaging, and a murder case exposing how little transparency we actually have when AI goes wrong. The pattern? Lots of companies are selling AI futures while quietly adjusting the present.
Here’s what caught my attention.
Why didn’t AI ‘join the workforce’ in 2025?
Sam Altman and Kevin Weil predicted 2025 would be the year AI agents started handling real work; booking travel, completing paperwork, and managing multi-step tasks autonomously. That didn’t happen. This matters because it shows how wide the gap remains between conference keynotes and actual capability. If you’ve been waiting for AI to transform your operations, you’ll keep waiting. Focus on what works today rather than what’s promised for next quarter.
AWS Quietly Raises GPU Prices 15% over the Weekend
Amazon Web Services raised prices on key GPU instances by roughly 15% on a weekend, with minimal notice. The p5e.48xlarge and p5en.48xlarge instances that power most serious ML workloads got more expensive across most regions. Cloud pricing has always been opaque. This is a reminder that your AI infrastructure costs can shift without warning. Build flexibility into your cloud strategy, or accept the bill.
Dell admits consumers don’t care about AI PCs
Dell dropped the AI messaging from its latest product presentations. After years of stamping “AI-powered” on everything, they’re acknowledging that consumers are tired of it. When hardware companies start walking back AI marketing, pay attention. The fatigue is real. Your customers may be feeling the same thing about your messaging.
Murder-suicide case shows OpenAI selectively hides data after users die
A disturbing case in which Stein-Erik Soelberg allegedly killed his mother after developing paranoia through ChatGPT conversations. OpenAI is refusing to release the chat logs. Whatever the facts of this case turn out to be, the legal and transparency questions are worth watching. When AI interactions contribute to harm, who has access to the data? The answer, apparently, is “not necessarily anyone.” As companies integrate AI tools into their operations, understand what you’re signing up for.
European Commission Issues Call for Evidence on Open Source
The European Commission is developing its European Open Digital Ecosystem Strategy to reduce dependence on non-EU software. If you sell software in Europe or work with European clients, this signals potential shifts in procurement policy. Open source vendors may find new opportunities. Proprietary software providers from outside the EU may face new friction.
UK Government Exempting Itself from Cyber Law Inspires Little Confidence
The UK’s new Cyber Security and Resilience Bill excludes both central and local government from its requirements. This, despite the fact that public sector organizations accounted for 40% of cyberattacks managed by the National Cyber Security Centre in 2020-2021. If you work with government agencies or handle public sector data, the double standard in cybersecurity requirements adds risk to your partnerships.
Donut Lab Launches Industry-Ready Solid-State Battery Technology
Donut Lab announced a production-ready solid-state battery hitting 400 Wh/kg energy density, with Verge Motorcycles planning to use it in their 2026 models. Solid-state batteries have been “coming soon” for years. This appears to be one of the first commercially viable options ready for actual manufacturing. Worth watching for anyone in the EV space.
Health Care Data Breach Affects Over 600k Patients, Illinois Agency Says
The Illinois Department of Human Services exposed data on 670,000 Medicaid/Medicare recipients through misconfigured privacy settings on public maps. The breach ran from 2021 to 2025. Four years. If you handle sensitive data, this is a reminder that breaches often aren’t sophisticated attacks but configuration mistakes that nobody catches.
Notion AI Data Exfiltration Vulnerability
Security researchers found that malicious prompts hidden in uploaded documents can exfiltrate sensitive data through Notion AI, even before users approve any AI interactions. If your organization uses Notion for confidential documents, this is the kind of risk that standard security protocols don’t catch. AI tools create new attack surfaces.
How Y Combinator Made It Smart to Trust Founders
A look at how Y Combinator shifted investor-founder dynamics, using Humble Bundle’s journey as an example. They went from a bootstrapped startup to a Ziff Davis acquisition, raising $274M for charity along the way. The larger point: treating founders as partners rather than replaceable assets has become a winning strategy in venture capital. Power dynamics in the startup ecosystem continue to shift.
That’s what I’m watching. What caught your attention this week?
-Eric
Go from AI overwhelmed to AI savvy professional
AI keeps coming up at work, but you still don’t get it?
Here’s what you get:
- Daily AI news that matters for your career - Filtered from 1000s of sources so you know what affects your industry.
- Step-by-step tutorials you can use immediately - Real prompts and workflows that solve actual business problems.
- New AI tools tested and reviewed - We try everything to deliver tools that drive real results.
- All in just 3 minutes a day