In partnership with
THE RUNDOWN
A few things worth your attention this week. Gallup crossed a threshold that's been coming for a while: half of US workers now use AI at work, and the productivity split inside teams is starting to show.
Meta set May 20 as the start date for 8,000 layoffs at a company that just posted $60 billion in profit. Earnings season heats up with Tesla, Intel, IBM, and Boeing reporting Wednesday and Thursday.
Let's get into it.
Quick Signals
Meta set May 20 as the start date for 8,000 layoffs. The internal memo confirms about 10% of Meta's global workforce, with more cuts planned for the second half of 2026. Meta earmarked $115 to $135 billion for AI infrastructure in 2026, so this is a reallocation (legacy budget out, AI budget in), not a financial squeeze.
Jobless claims dropped to 207,000, the biggest one-week drop since February. Claims fell 11,000 versus expectations of 215,000, with the 4-week moving average at 209,750. Layoffs are staying low even as the hiring narrative feels uncertain, which is a more useful signal than the headlines suggest.
Industrial production unexpectedly fell 0.5% in March. Manufacturing output slipped 0.1% and capacity utilization dropped to 75.7%, below its long-run average. If you're in industrials, logistics, or supply chain, the factory floor is softer than the stock market right now.
JPMorgan beat Q1 earnings but lowered full-year guidance, and this week four more heavyweights report. The bank reported EPS of $5.94 versus $5.45 expected, then cut full-year net interest income guidance from $104.5B to $103B. Tesla, IBM, Intel, and Boeing all report Wednesday and Thursday, and Intel's guidance is the one to watch if you work anywhere near semis or AI infrastructure.
Anthropic released Claude Opus 4.7 on Thursday. The new model adds self-verification, a 1-million token context window, and an 87.6% SWE-bench score, same price as 4.6. Available in Claude, the API, AWS Bedrock, Google Vertex, and Microsoft Foundry.
A Word from Our Partner
Rex Gelb spent a decade building HubSpot's paid engine. Now he's showing founders exactly how to do it.
On April 27th, get the framework to structure, launch, and scale paid media that drives pipeline, not just traffic. 20 minutes. Live Q&A. Free.
The Big Story
Half of US workers now use AI at work. That changes the math.
Gallup surveyed 23,717 US employees in February and just published the results: 50% of employees say they use AI in their role at least a few times a year, up from 46% last quarter and 21% in 2023. 13% use it daily and 28% use it a few times a week or more, both all-time highs.
The more useful number is buried underneath. 65% of AI users say it's improved their productivity and 16% call the impact "extremely positive," while only 9% say it's had a negative effect on their work.
Here's the quiet part. 91% of companies have deployed AI somewhere in the business but only 29% report significant ROI, which means the gap is no longer between companies that have AI and companies that don't.
The split is inside the same company. Among employees who have AI tools available but don't use them, 46% say they prefer their current workflow, and when half a team saves 40 to 60 minutes a day and the other half doesn't, that gap gets visible to the boss within a quarter.
Why it matters: If you're not in the weekly-user tier yet, getting there is cheaper than most professional upgrades and faster than a cert. The specific prompt to start with today is in Try This below.
Making Moves
The Class of 2026 is walking into the toughest entry-level market in 37 years. Except in three places.
Employers are projecting a 1.6% hiring increase for the Class of 2026 versus 2025, per NACE. More than 60% of the class is pessimistic about their career prospects, and more than half of employers describe the entry-level market as "poor" or "fair."
Underneath that, the real story. New workforce entrants made up 10.6% of the unemployed in February 2026, higher than at any point in the Great Recession, while finance and information services (the two industries that used to absorb the bulk of new grads) have shed an average of 9,000 jobs a month since 2023.
Three lanes are actually expanding. Nursing grads earned a $70,000 median starting salary, 16.7% above expectations, with 31.8% locking in jobs before graduation.
Computer science grads are starting at $81,535, up nearly 7% year over year. Healthcare broadly is projected to generate 1.8 million new positions a year through 2030, with nurse practitioner roles growing 40% by 2034.
The other divider inside the class: internship experience. Grads with work experience were hired at 81.6% within three months, compared to 40.7% for those without, and applications per role are up 26% while postings are down 16%.
Why it matters: If you're entering this market, or advising someone who is, the highest-leverage move is a lane switch toward healthcare, nursing, or technical engineering, plus any structured experience that signals applied skill (bootcamps, freelance work, research assistantships count).
Try This Out
The one prompt that will get you into the AI-at-work half this week.
Open Claude or ChatGPT, paste this prompt in, save it in your notes app, and use it on the next five emails or updates you write. Within three uses, you'll know whether AI belongs in your weekly flow.
You are my writing assistant. I'll give you rough notes, bullets, or a
stream-of-consciousness draft, and I want you to turn it into a clean,
professional message.
Rules:
- Match my tone: direct, warm, no fluff, no corporate hedging
- Keep it as short as possible while covering everything I wrote
- End with a clear ask, next step, or question
- Give me three subject line options at the top, each under 8 words
First, ask me two clarifying questions before you draft (who the audience is
and what outcome I want), unless the answers are obvious from what I pasted.
Here's the rough version:
[PASTE YOUR ROUGH NOTES]
Two things to actually notice as you use it. First, the clarifying questions force you to be specific about the audience and outcome, which is usually where bad first drafts go wrong. Second, the three subject-line options exist so you stop wasting five minutes picking one, you pick in ten seconds and move on.
A second prompt worth stealing: the same skeleton, but swap "writing assistant" for "meeting prep assistant" and "rough notes" for "invite + attendee list." You'll walk into meetings prepared in two minutes instead of twenty.
Predict This
When will 60% of US workers use AI at work?
By end of 2026
First half of 2027
Second half of 2027 or later
Gallup's 50% milestone took about three years from the 21% baseline. The next 10 points will likely come faster or slower depending on how many companies finally fund AI training (56% of the global workforce still hasn't received any).
Worth Reading
Seven days of meditation rewires the brain, UCSD / Communications Biology
8,000-year-old Mesopotamian pots encoded early math for sharing land and crops, 2026 in archaeology
