Stanford's Canaries Dashboard Just Killed the New-Grad Funnel
Stanford's Canaries Dashboard confirms 22-25 SWE employment is down ~20% while 30+ engineers keep growing. How to re-target your pipeline now.
On June 10, 2026, Stanford Digital Economy Lab and ADP Research launched the Canaries Dashboard at indicators.stanford.edu, giving talent leaders a near-real-time, monthly view of how AI is reshaping employment by occupation. The headline number is brutal for anyone still running a new-grad inbound funnel: 22-25 year-old software developers are down roughly 20% from their late-2022 peak, while engineers 30+ have grown 6-12% over the same window. If your 2026 plan still assumes a healthy junior pipeline feeding mid-level seats in 18 months, the dashboard is telling you that math is broken.
This is not a doom story. It is a re-targeting story. The supply that is growing, and the demand that is paying premiums, both sit in the same cohort: experienced engineers. Most sourcing stacks are pointed at the cohort that is shrinking.
What the Canaries Dashboard actually measures
The Canaries Dashboard is one of three dashboards in Stanford's new AI Economic Indicators platform, built in collaboration with ADP Research and grounded in the "Canaries in the Coal Mine" paper by Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen. The balanced sample at launch covers 25,000 firms over the five years ending April 2026, employing 4.6 million workers matched to 730+ occupation codes. It updates monthly. That cadence matters: every other labor signal sourcers rely on (BLS, JOLTS, university placement reports) lags by quarters.
Erik Brynjolfsson framed the launch bluntly: "We cannot afford to rely on anecdotes or lagging indicators of AI's effects." ADP Chief Economist Dr. Nela Richardson called it "a near real-time view of how AI is reshaping work."
The age buckets in the underlying paper are worth memorizing because they will become the new shorthand in sourcing conversations: Early Career 1 (22-25), Early Career 2 (26-30), Developing (31-34), Mid-Career 1 (35-40), Mid-Career 2 (41-49), and Senior (50+). The collapse is concentrated in Early Career 1. Early Career 2 has drifted slightly down. Everything 31+ is flat to growing.
The mechanism: automation, not exposure
The most important finding in the paper is one most coverage missed. Occupations where AI automates work see declining youth employment. Occupations where AI augments workers saw employment grow for workers of all ages. The exposure category alone is not predictive. The automation-vs-augmentation ratio is.
For sourcers, this means "AI-exposed = avoid the role" is the wrong heuristic. The right one is: which skill stacks indicate the candidate sits on the augmentation side of that ratio? AI-tooling fluency, code review at scale, system design, mentorship of junior engineers using copilots. Those are the markers that the role is getting amplified, not eaten.
Why the 30+ cohort is the only pipeline that math still works
Two forces are squeezing the same cohort from opposite sides.
On the supply side, the Canaries data shows 30+ engineers are the only group whose employment trends are stable or growing. A 2025 LeadDev survey found 54% of engineering leaders plan to hire fewer juniors specifically because AI copilots let seniors handle more. That is the augmentation effect showing up in budget decisions. CS grads now face a 6.1% unemployment rate and computer engineers 7.5%, among the highest of any major. Entry-level SWE job postings dropped 28% from 2022 peaks and have not recovered through 2026.
On the demand side, demand for AI/ML skills among junior candidates fell far more steeply than for seniors, because AI projects typically demand deep expertise. The seniors are not just safer. They are commanding premiums.
number: 237,000
label: US-based engineers carrying Senior, Staff, or Principal SWE titles in Refolk's index
note: Concentrated in San Francisco, Seattle, LA, and NYC. Top employers include Google, Datadog, and Starburst.