In partnership with
In Today’s Issue:
🤑 43% of CEOs plan to cut junior roles as AI reshapes hiring
💵 Former Microsoft VP calls Copilot strategy a failure
📈 Utah's $100B+ datacenter faces atomic-scale heat pushback
📉 Mistral CEO gives Europe a two-year AI ultimatum
✨ And more AI goodness…
⚡ The Signal
The labor market is splitting along an age line, and AI is the blade.
Companies are not cutting headcount because times are bad. They are cutting junior headcount because AI makes entry-level tasks automatable at a speed that makes hiring graduates for them economically irrational.
The Oliver Wyman data, the New York Fed numbers, and Microsoft's own Copilot retreat all point to the same structural shift: the organizations being built for the AI era are flatter, older, and smaller.
The uncomfortable question nobody in a boardroom wants to answer is: who trains the next generation of mid-level talent when the entry-level rung no longer exists?
All the best,
Kim Isenberg
⚡ Utah Datacenter Faces Heat Backlash
A proposed 9 GW datacenter campus in Utah’s Hansel Valley is drawing scrutiny after physicist Rob Davies estimated it could release heat equal to 23 Hiroshima-sized atomic bombs per day, potentially raising local temperatures and stressing a fragile desert ecosystem. Critics dispute parts of the claim, but the larger tension is whether the Stratos project can survive environmental pushback, community opposition, and financing needs that could climb past $100 billion.
👉 tl;dr: Utah’s proposed Stratos datacenter could become one of the world’s biggest AI infrastructure bets, but its heat, cost, and environmental risks may decide whether it ever gets built.
🤖 Microsoft Predicts White-Collar AI Reckoning
Microsoft AI CEO Mustafa Suleyman says AI could automate most computer-based professional work within 18 months, putting fields like law, accounting, marketing, coding, and project management under pressure. Fortune’s article balances that stark forecast against mixed real-world evidence so far: modest productivity gains, some failed returns, and early AI-linked layoffs, suggesting the threat is serious but not yet fully proven.
👉 tl;dr: Microsoft’s AI chief warns that white-collar automation may arrive fast, though current evidence shows disruption is still uneven and contested.
⚠️ Europe’s AI Clock Ticks Faster
Mistral CEO Arthur Mensch warned French lawmakers that Europe has just two years to build serious AI infrastructure or risk dependence on US giants for chips, energy, compute, and models. His starkest claim is geopolitical: without control over the systems that turn “electrons into tokens,” Europe could become an AI “vassal state” with little leverage over American tech power.
👉 tl;dr: Mistral’s CEO says Europe must urgently build its own AI infrastructure or become locked into dependence on US tech giants.
Use AI to audit your own job for automation risk before your boss does.
Why it helps: The Oliver Wyman survey shows CEOs are quietly mapping which roles can be replaced. Running the same analysis on yourself lets you spot the gap between what you do today and what will still matter in 18 months, so you can shift your time toward the parts of your job that AI handles poorly: judgment calls, cross-team coordination, and ambiguous problems.
Try this: Paste your current job description into Claude or ChatGPT and ask: "Which of these tasks could an AI assistant handle today at 80% quality? Which require human judgment that AI can't replicate? Rank them." Then spend next week deliberately building skills in the second column.
🎬 Watch This
Bloomberg's Odd Lots podcast digs into the most viral chart in AI right now: METR's benchmark showing AI model capabilities going up and to the right at an accelerating pace. The episode features METR's president Chris Painter explaining how the organization measures whether AI can handle autonomous, complex tasks, and why that specific capability is what keeps researchers up at night. Worth watching alongside today's Graph of the Day, which shows a similar curve for autonomous protein folding research.
“Happiness isn’t fully determined by circumstance, but by repeatedly choosing how we interpret, focus on, and respond to life.”
— Jack Altman
AI Is Gutting the Junior Job Pipeline
The Takeaway
👉 The share of CEOs planning to cut junior roles has doubled to 43%, while only 17% are shifting hiring toward entry-level positions, according to a new Oliver Wyman global survey.
👉 The job market for 22-to-27-year-olds hit its worst level since the pandemic, with Fed Chair Jerome Powell acknowledging AI is partly to blame as companies automate work that would have gone to recent graduates.
👉 74% of CEOs are either freezing or reducing headcount overall, up from 67% last year, with the most aggressive cuts in tech, media, and telecommunications.
👉 A small contrarian group of advanced AI adopters actually see entry-level talent becoming more valuable, not less, suggesting the relationship between AI maturity and junior hiring may be more complex than the headline trend.
The numbers paint an uncomfortable picture for anyone entering the workforce in 2026. An Oliver Wyman survey of global CEOs shows that 43% now plan to reduce junior roles over the next one to two years, more than double the 17% who said the same last year. At the same time, companies are pivoting toward mid-level hires. Roughly 30% of executives are shifting recruitment to experienced workers, up from just 10% a year ago.
The reason is straightforward: AI in its current form is best at automating exactly the tasks that early-career employees perform.
The consequences are already showing up in federal data. The New York Fed reported that the job market for 22-to-27-year-olds deteriorated noticeably in the first quarter of the year. Fed Chair Jerome Powell later confirmed that AI is a contributing factor, explaining that companies are choosing to automate rather than hire graduates for tasks that used to serve as the first rung of the corporate ladder. By year-end, these workers faced the harshest employment conditions since the peak of the pandemic.
The deeper risk goes beyond the current cohort. Less hiring at the entry level means fewer opportunities for on-the-job training, which hollows out the talent pipeline that companies depend on for future mid-level and senior hires. Executives who are most advanced in AI adoption showed a slightly higher willingness to invest in junior talent, hinting that the relationship between automation maturity and workforce structure may evolve. But the dominant trend is clear. The CEOs with the longest planning horizons are the ones most aggressively cutting headcount, viewing leaner organizations not as a temporary cost measure but as a permanent destination.
Why it matters: Companies are not just reducing entry-level hiring as a cost measure. They are structurally rewriting how organizations are staffed, which threatens the career pipeline that produces the mid-level and senior talent they will need in five years.
Sources:
🔗 https://gizmodo.com/the-young-are-being-battered-by-ai-as-hiring-shifts-to-older-workers-2000759608
🔗 https://fortune.com/2026/05/01/automating-gen-z-entry-level-jobs-could-backfire-mit-ai-researcher-andrew-mcafee-talent-pipelines-at-risk/
🔗 https://fortune.com/2026/04/06/ai-tech-displacement-effect-gen-z-16000-jobs-per-month/
The IT strategy every team needs for 2026
2026 will redefine IT as a strategic driver of global growth. Automation, AI-driven support, unified platforms, and zero-trust security are becoming standard, especially for distributed teams. This toolkit helps IT and HR leaders assess readiness, define goals, and build a scalable, audit-ready IT strategy for the year ahead. Learn what’s changing and how to prepare.
The chart: OpenAI's Codex autonomously iterated on SimplexFold, an AlphaFold2-style protein folding model, across 127 experiment runs. The best validation C-alpha lDDT score climbed from around 0.20 to 0.4311, with a sharp acceleration after run 100 as the system found increasingly effective configurations on its own.
The lesson: AI systems can now run scientific research loops autonomously. Codex did not just write code here; it designed experiments, evaluated results, and improved a protein folding model without human intervention between runs. That is a qualitative shift from AI-as-tool to AI-as-researcher.
The caveat: A validation lDDT of 0.43 is still well below production-grade protein structure prediction. The chart shows the method works in principle, but the absolute accuracy gap to AlphaFold3-class models remains large. Autonomous research is impressive; autonomous good research is a different bar.
Microsoft's AI Strategy Under Fire from the Inside
⚡ Bottom line: A former Microsoft VP who held senior AI roles at Google and Meta says Microsoft missed the AI wave entirely.
💡 Why it matters: Only 3.3% of Microsoft 365 users pay for Copilot, exposing a widening gap between AI spending and real adoption.
🔎 What it means: Microsoft is already retreating from aggressive Copilot integration across Windows 11, suggesting the company knows its approach backfired.
Mat Velloso spent years inside Microsoft's leadership before moving to senior AI product roles at Google DeepMind and Meta. His verdict on his former employer is blunt: Microsoft missed the AI wave the same way it missed the internet and mobile. Velloso points to Copilot as the central failure, noting that out of 450 million Microsoft 365 users, only about 15 million are paying for Copilot—a 3.3% conversion rate that looks alarming against the company's estimated $37.5 billion in quarterly AI spending.
The hardware story is equally damaging. Microsoft pushed OEMs to build laptops with dedicated Neural Processing Units (NPUs) for AI features, but Velloso says manufacturers invested heavily only to find that nobody cares because no valuable use case was built for those chips in Windows or Office. Meanwhile, Microsoft is quietly rolling back Copilot integration from Notepad, Photos, Snipping Tool, and Widgets after months of user backlash. The Copilot branding in Notepad has been replaced with a generic "Writing Tools" label.
The external pressure compounds the internal confusion. OpenAI, Microsoft's closest AI partner, is building its own enterprise infrastructure that directly competes with Microsoft's corporate business. Velloso's advice is a full internal reset. Whether Redmond listens or continues playing defense against its own partners remains the open question.
Are you tracking agent views on your docs?
AI agents already outnumber human visitors to your docs — now you can track them.
