Engineering Hiring Fell Less Than Overall Tech Hiring

SignalFire’s 2026 data describes a smaller decline in engineering hiring than in total large-tech hiring: 11% versus 25% compared with 2019. Engineers accounted for 55% of new hires, up from 46%. These figures do not establish that AI caused the difference or that engineering jobs are safe from automation.

Use the hiring figures as context for your own staffing decision. They do not establish that every company should add engineers or that a smaller decline means protection from AI-related change.

Are engineering jobs safe from AI? What the new data actually says

Engineering hiring declined less than overall hiring in the reported dataset. While total hiring across large tech companies fell 25% versus 2019, engineering hiring fell only 11%, roughly half the decline. That gap is the headline finding of SignalFire’s State of Talent Report, as reported by TechCrunch on June 24, 2026.

The report tracks careers across an enormous base, more than 80 million companies, and classifies twelve firms as “Tech Majors”: Alphabet, Meta, Apple, Amazon, Microsoft, Netflix, Nvidia, Tesla, Uber, Airbnb, Block, and Stripe. These are the most AI-forward companies on earth. They are the ones shipping the coding models. And they are the ones hiring engineers at a higher rate than they hire anyone else.

The composition shift is even clearer than the survival rate:

Engineers did not just hold their ground. They became a bigger slice of a smaller pie. The lean 2026 company is not simply smaller. It is a senior-heavy engineering core with the surrounding support functions stripped out.

Even early-stage startups grew their engineering headcount

The most on-brand finding for founders is the sharpest one. Early-stage startups brought on 7% more engineers in 2025 than in 2019, a net increase against a market that contracted everywhere else. The smallest, leanest, most budget-constrained companies, the ones with no room for a wrong bet, chose to add engineers.

That matters because early-stage companies are forced to be honest about leverage. A 400-person company can carry a role that does not pull its weight. A six-person company cannot. When that kind of company grows engineering against the trend, it is telling you where the return is. We made the same observation in our playbook on a founder’s first five hires: the founding engineer is the position investors and data agree on first, before any commercial role.

A possible explanation: greater demand for software

One possible explanation is that cheaper software production increases demand for software. This resembles the Jevons-paradox hypothesis, but the hiring figures alone do not test it. Product demand, company budgets, and the composition of the sample may also affect hiring.

The people closest to the work describe exactly this. Asher Bantock, SignalFire’s head of research, says engineers are “suddenly a lot more productive, and there’s endless work for them to do.” Jensen Huang, Nvidia’s CEO, says “software engineers are busier than ever” even after the company adopted agentic AI that writes code “near instantaneously.”

The intuition that AI would shrink engineering assumed the amount of software a company wants is fixed. It is not. Every startup has a backlog longer than its runway. When each engineer can do more, the rational move is not to hire fewer of them. It is to point more of them at a backlog that was never the bottleneck. The bottleneck was always the throughput, and AI raised the ceiling on what a strong team can ship.

The honest counter-argument: entry-level is collapsing, the discipline isn’t

Here is the part the optimistic version skips, and you need it to make a good decision. The resilience is concentrated in experienced engineers, not new graduates. The discipline is growing. The traditional on-ramp into it is contracting hard.

The numbers are stark:

• New grads are just about 7% of Big Tech hires, down more than 50% from 2019, per SignalFire’s 2025 Tech Talent Report. At startups they are under 6% of hires.
• Since 2021, the average age of technical hires rose about three years as companies stopped investing in training juniors.
• Stanford’s Digital Economy Lab found that employment for software developers aged 22 to 25 fell roughly 20% since late 2024, even as headcount for developers 30 and older at the same firms grew.

There is a real debate about how far this goes. Anthropic CEO Dario Amodei has warned that AI could wipe out half of entry-level white-collar jobs. Anthropic’s own head of economics, Peter McCrory, is more measured, noting “there’s at least no larger material difference in unemployment rates” between AI-exposed and physical-interaction roles so far. Both can be true: the on-ramp is narrowing while the broad collapse some predicted has not arrived.

The founder takeaway is not “junior engineers are worthless.” It is that the bar and the role have shifted. The work that AI now does, generating boilerplate and first-draft code, is exactly the work junior engineers used to cut their teeth on. So you cannot hire on credentials and a clean resume anymore. You have to hire on demonstrated judgment.

Engineers are absorbing the management layer

As the support structure around engineering thins, the work does not vanish. It moves onto the engineers. SignalFire’s data shows non-technical functions (recruiting, parts of product, sales support) shrinking while the engineering core grows. The coordination layer that used to sit between strategy and code is getting compressed.

One analysis of the report pegs middle-management cuts at 41%, the layer that once “translated business strategy into engineering roadmaps.” As that layer thins, engineers inherit product decisions, customer context, and business tradeoffs. The modern senior engineer is not just shipping features. They are deciding which features to ship and why.

This reshapes what you are actually hiring for. The engineer who can write code is now table stakes, because AI writes code too. The engineer who can own a problem end to end, reason about tradeoffs, talk to a customer, and decide what is worth building, that is the scarce and resilient profile. Output is commoditized. Judgment is not.

What this means for founders: keep hiring engineers, and hire for judgment

The data points to three moves, in order.

1. Keep investing in engineering headcount. It is the function the market is protecting, and at early stage it is the function still growing. This is where leverage compounds. Cutting it to “wait for AI” gets the trade backwards.
2. Hire for judgment, not just output. When AI handles boilerplate and engineers absorb product and business decisions, your differentiator is engineering judgment: system design, tradeoff reasoning, ownership, communication. Assess these skills directly rather than assume that producing working code demonstrates them.
3. Run a process that actually surfaces judgment. If everyone’s coding is AI-augmented, a take-home that measures whether someone can produce working code tells you almost nothing. Your screening has to test the human layer, the design reasoning and the debugging of unfamiliar systems, not typing speed.

That third move is where most lean teams fall down. They believe engineering is the bet, they fund it, and then they evaluate candidates with a process built for 2018, one that rewards exactly the skills AI just commoditized.

How to hire resilient engineers well

The data makes the case to keep hiring engineers. The hard part is hiring them well when AI has flattened the old signals. Kit is the AI-native ATS built for startups doing exactly that, and a structured process maps cleanly onto the shift the data describes.

Define the role around judgment, not a framework checklist. Write the job posting for ownership and tradeoff reasoning, the non-automatable layer the market now rewards, instead of a list of libraries. A candidate who lists ten frameworks and a candidate who can explain one hard architectural decision are not the same hire, and the second one is the one the data favors. Our guides on how to hire a founding engineer and how to hire a full-stack engineer go deeper on framing each role.

Run a repeatable, structured pipeline. A lean team without a recruiter still needs to evaluate every candidate the same way, or the decision drifts toward whoever interviewed last. Process templates give you a consistent set of stages so a three-person company hires with the rigor of a forty-person one.

Use code assignments that test reasoning, not output. Since AI now produces clean first-draft code on demand, a good assignment probes the things it cannot fake: debugging an unfamiliar codebase, defending a design tradeoff, explaining why one approach beats another. That is precisely the judgment layer engineers are absorbing as management thins.

Score on demonstrated signal, not credentials. Anchored scorecards and structured team reviews keep the decision on evidence of reasoning rather than the resume, which is exactly the shift the entry-level collapse forces on you. We cover this method in detail in skills-based hiring with structured scorecards, and the related challenge of screening for AI over-reliance when every candidate codes with an assistant.

The bottom line

Engineering is the most AI-resilient function in tech, and the smallest companies are growing it fastest. AI did not gut engineering. It made each engineer more valuable and pushed product and business judgment onto them, while quietly closing the junior on-ramp. The discipline is the bet worth making. The challenge is hiring for the judgment that now defines it, when AI has erased the easy signals.

That is the work. The data says engineering is the bet worth making; a structured, signal-based process is how you make it well. If you are building a hiring pipeline for engineers this year, start a free Kit trial and set it up around judgment from day one.