Why does the job market feel broken even after 172,000 jobs were added?

Last Friday the Bureau of Labor Statistics reported that 172,000 jobs were added in May. That was more than double what economists expected. The unemployment rate held steady at 4.3%. The headlines called it a strong report. Some economists went further and called it the end of the hiring recession.

I read the full report over the weekend. Then the Indeed Hiring Lab analysis published Monday. Then the Center for American Progress breakdown. Both readings are defensible. The economy is adding jobs, and people are still struggling to connect with work. Both things are true at the same time, and the gap between them is exactly where everyone reading this newsletter does their work.

By the numbers, in 20 seconds

• 172,000 jobs added in May, double the 85,000 economists expected. Unemployment held at 4.3%.
• Long-term unemployment is up 524,000 in a year. 27.5% of all unemployed people have now been out of work 27+ weeks.
• Entry-level tech hiring is down 73.4% while AI engineering postings grew 143% year over year. The market is splitting, not collapsing.
• DFW is running against the grain. Metro unemployment fell from 4.2% in January to 3.8% in April, led by professional and business services, up 15,500 jobs over the year.
• Workforce Pell goes live July 1. Short-term programs nationwide must prove 70% completion and 70% job placement to qualify for federal aid.
• The first HB8 downward adjustments hit the October payment. Texas colleges have one summer to find out where they stand.

The two realities

The headline says 172,000 jobs. The details say something more complicated.

First, the genuinely good news. May's gain came in at double the consensus forecast. March and April were revised up by a combined 93,000 jobs, so the spring labor market was stronger than first reported. The broadest underutilization measure, U-6, actually edged down. If you only watch the top lines, this was the best report in months.

Now the other reality. The unemployment rate has been stuck between 4.3% and 4.5% since July 2025, and what is happening inside that stability is the story. The number of long-term unemployed workers, people out of work 27 weeks or more, rose by 524,000 over the past year to 2 million. They now make up 27.5% of all unemployed people, up from 20.4% a year ago. People who lose jobs are taking far longer to find new ones.

Indeed's research team calls it a "low-hire, low-fire" equilibrium. Companies are not laying off in large numbers. They are also not hiring at the pace of two years ago. The Center for American Progress adds that the share of Americans who want a job but are not counted in the labor force remains well above pre-pandemic levels. Stable is not the same as strong for the people looking right now.

A 7-point jump in 12 months. The hiring side of the market has stabilized. The re-employment side has not.

The AI factor

AI is creating jobs and eliminating them at the same time. That is what makes this moment so hard to read.

While the BLS report showed broad gains in healthcare, local government and leisure and hospitality, the technology sector tells a different story. Over 100,000 tech jobs have been cut in 2026 so far. Oracle announced plans to cut 20,000 to 30,000 roles to redirect $8 to $10 billion toward AI infrastructure. Block cut 4,000 roles with its CEO explicitly citing AI efficiency. Meta cut roughly 8,000 and reassigned 7,000 more to AI teams.

At the same time, LinkedIn ranked AI engineer as the fastest-growing job title in the US for 2026. AI and machine learning postings grew 143% year over year, and PwC's AI Jobs Barometer finds workers with AI skills command wage premiums up to 56%. MIT Technology Review's read last week was sober. Hard for new graduates (recent-grad unemployment near 5.6%, well above the overall rate), but no aggregate AI jobs collapse. What has materialized is a reshuffling. Ravio's compensation data shows entry-level tech hiring down 73.4% across the industry.

The labor market is not collapsing. It is splitting. One track is growing. One track is shrinking. The workers caught between the two tracks are the ones the headline number does not capture.

The compliance clock

Here is where the national story becomes your story. A split labor market would be just commentary if public funding still flowed on enrollment. It does not anymore, and two clocks are running.

Workforce Pell goes live July 1, 20 days from now. Final federal rules landed May 19. For the first time, students in 8-to-15-week programs can use Pell Grants, but only in programs that clear hard outcome bars: at least 70% completion, at least 70% job placement within 180 days and an earnings test. Governors and workforce boards must designate the high-demand occupations these programs serve. Read that against this jobs report. The sectors that qualify as high-demand are exactly the sectors adding jobs, healthcare first. The colleges that can map their short-term programs to growing regional occupations, with evidence, will capture this money. The ones that cannot will watch it go elsewhere.

The first HB8 downward adjustments hit Texas colleges in the October payment. HB8 funding is dynamic now. FY25 overperformance was added to the February payment this year. FY25 underperformance comes out of the October one. Every Texas community college has one summer to know which of its programs are producing fundable outcomes, which credentials count as credentials of value and where the misclassifications are hiding. The legislature is also refining the credential-of-value definition with more precise labor market data, so the target is moving while the clock runs.

What our data shows: DFW is running against the national grain

National numbers hide regional stories, so each issue now closes the loop with the region we track most closely. More than 52,000 active postings across 47,000+ North Texas employers. Right now, DFW is diverging from the national story in three ways worth knowing.

First, while the national conversation is about rising long-term unemployment, DFW unemployment has been falling. From 4.2% in January to 3.8% in April, below the comparable national rate. The Fort Worth-Arlington-Grapevine side of the metro is running even tighter at 3.7%. Second, the metro's growth leader is professional and business services, up 15,500 jobs over the year, which cuts directly against the national white-collar hiring freeze narrative. The freeze is real nationally. It is not what DFW employers are doing in aggregate. Third, leisure and hospitality slipped locally even as it led national gains in May, a reminder that your metro and the headline can move in opposite directions in the same month.

Five of DFW's ten major sectors added jobs over the year through April. If you lead a college or an EDC here, the question is not whether the region is growing. It is whether your programs and your pitch are pointed at the five that are. If you want this view for your own county or metro, that is what the office hours below are for.

Metro figures: BLS metropolitan area employment and unemployment data through April 2026, the most recent available. Metro data lags the national report by about one month.

What this means by audience

For EDC directors. One regional unemployment rate no longer answers the talent question. Healthcare is hiring. Local government is hiring. Financial services lost jobs in May. Tech is shedding experienced workers while posting AI roles that barely existed two years ago. When a prospect asks about your talent picture, the EDC that can show which local sectors are growing, which are contracting and what the education pipeline produces for each will win the conversation. Sector-level, current, local. That is the new bar.

For college and university leaders. The 73.4% decline in entry-level tech hiring is the most urgent number in this report. Your computer science and business graduates are walking into a market where the roles they trained for are disappearing while roles they were not trained for multiply. The institutions adapting fastest are co-designing programs with employers, models like BILT where companies sit on the curriculum committee, and embedding AI fluency into existing programs. With Workforce Pell live in 20 days, your short-term programs now need outcome evidence, not just enrollment.

For employers. Low-hire, low-fire feels stable. It is not. Your experienced people are staying because they see no better options. Indeed's analysis warned this equilibrium holds only until something pushes on it. The companies building apprenticeship and co-op pipelines with local colleges now are the ones with a bench when the market shifts.

For workforce boards. The 524,000 rise in long-term unemployment is a direct signal about your caseload. People in your programs are taking longer to land. That is market structure, not program quality. Training built for an economy hiring broadly needs recalibration for one hiring selectively, and your new role in designating Workforce Pell occupations makes your demand data more consequential than it has ever been.

Three things forward-looking leaders are doing right now

1. They are reading the jobs data at the sector level, not the headline level. May's gains came in healthcare, local government and leisure and hospitality. Financial activities lost jobs. The leaders I talk to are matching their regional program output to the sectors actually growing in their metro, not to the national headline. Annual reports cannot capture a market moving this fast.
2. They are building AI skills into existing programs rather than launching standalone AI degrees. AI skills command up to a 56% wage premium, but most students do not need an AI degree. They need AI fluency in their field. A nursing student who understands AI triage tools. A supply chain manager using AI demand forecasting. The institutions adding AI modules to healthcare, business and engineering programs are moving faster than the ones building new departments.
3. They are treating the long-term unemployed as a pipeline, not a problem. Many of the 2 million long-term unemployed have experience and education. What they lack is a bridge to the roles forming now. Employers and boards that treat this group as a reskilling opportunity are getting ahead of the curve, and in Texas the Skills Development Fund can offset up to $500,000 per employer for exactly this kind of customized retraining.

What to watch: Indeed Hiring Lab's labor market analysis

Indeed's Hiring Lab published their May analysis Monday under the title "One Strong Headline, but Two Realities." It is the clearest breakdown I have read of why the headline number and the lived experience of job seekers diverge right now. What makes it valuable for workforce leaders. They publish at the metro level, so you can see whether your region matches the national picture or diverges from it. Free, monthly, at hiringlab.org.

If your organization is doing original labor market research that connects to workforce outcomes, I would welcome featuring your work. Reply to this newsletter or reach out on LinkedIn.

Data source of the issue: Center for American Progress jobs analysis

CAP publishes a detailed analysis of every monthly BLS report that goes well past the headline, including the underutilization measures and the people who want work but are not counted in the official rate. It publishes the same day as the jobs report. A strong complement to the raw BLS data for board analysts and institutional researchers.

Workforce data office hours, Fridays in June

I am holding free 30-minute sessions for EDC directors, institutional researchers and workforce board analysts who want to see what real-time regional workforce intelligence looks like for their area. No pitch. Just your region's data on the screen. Reply to this newsletter or DM me on LinkedIn to reserve a Friday slot.

P.S. If you run short-term workforce programs anywhere in the country, the July 1 Workforce Pell date is the one to circle. I am putting together a plain-English eligibility walkthrough. Reply with "Pell" and I will make sure you get it first.