More AI. More Opportunity. The Mission Engineering Rebound.
The evidence does not support a simple IT jobs apocalypse: BLS projections, observed tech employment, and GitHub activity point to changing work and expanding systems demand—not a guaranteed headcount collapse.
More AI. More opportunity. More missions to engineer.
The simplest version of the AI jobs-apocalypse story says that as software becomes cheaper to build, the people who build and operate it become unnecessary. The data do not establish that outcome. The better hypothesis is a demand rebound: lower software-production costs make vastly more systems worth creating, testing, securing and operating.
Six signals, with their evidence types kept separate
| Signal | Value | Evidence type |
|---|---|---|
| U.S. computer systems design and related services employment, September 2026 | 2.355 million | Observed payroll employment BLS |
| Software developer employment, 2024–2034 | +15.8%; +267,700 jobs | BLS occupational projection BLS |
| Information security analyst employment, 2024–2034 | +28.5% | BLS occupational projection BLS |
| GitHub developer accounts in Octoverse 2025 | 180 million+ | Platform accounts, not jobs GitHub |
| New GitHub pull requests, September 2026 | 23.0 million; +186.5% YoY | Third-party estimate, includes possible automation Tracker |
| New GitHub repositories, September 2026 | 11.9 million; +109.1% YoY | Third-party estimate Tracker |
What the evidence actually proves
The U.S. industry employed roughly 2.36 million people in September, down slightly from August. That is not evidence of a jobs boom, but neither is it evidence of extinction. BLS projects continued growth for developers and security analysts over the coming decade, while GitHub measures a large and active ecosystem. These are different populations and different time windows. We do not treat them as one observed time series or interpret a GitHub account as a paid job.
The Frontier scenario: demand can eat the efficiency dividend
Consider an explicitly synthetic, illustrative enterprise scenario. An organization operates 500 systems today. AI reduces direct implementation and operating labor per system by 90%, but makes 5,000 systems economically worthwhile. Total direct work units stay at 500 (500 × 1 versus 5,000 × 0.1)—before accounting for the integration, governance, security and verification burden of a tenfold system estate.
This is not a forecast. It is the Jevons/rebound mechanism made testable: the employment outcome depends on how much new demand cheaper systems unlock, and how much mission oversight they require.
The emerging scarce capability is not typing code. It is Mission Engineering: defining the mission, directing automated builders, proving that systems work, and owning the outcomes.
AI produces the code. Mission Engineers produce the results.
Methodology and cautions
Observed employment, occupational forecasts, GitHub platform counts, third-party activity estimates and synthetic scenarios are explicitly labeled. The branded graphic is an editorial illustration, not a verified continuous historical chart. Its intermediate index points should not be interpreted as observations. The source-backed data object is stored with this Data Insight.