TrueUp is tracking 67,000 open software engineering roles at tech companies right now, up 30% since January. That number doubled from the mid-2023 trough. Amit Taylor, TrueUp's founder, said it plainly in early April: "A lot of the 'AI is replacing engineers' narrative isn't grounded in job posting data, at least not so far." That's not a press release. That's a database.
So yes, the boom is real. But the distribution of that boom is where things get complicated, and where I think most of the commentary gets it wrong.
Two Markets Wearing the Same Job Title
Median base salary for software engineers at venture-backed startups hit $200,000 this year, a 25% jump from 2022 according to Levels.fyi data from March 31. Seed-stage startups are now offering new CS grads up to $300,000 base. Chris Vasquez, CEO of Quantum, told the WSJ he'd never seen that before at seed companies. OpenAI averaged $1.5 million in stock-based compensation across its 4,000 employees last year.
Meanwhile, the average starting salary for the CS class of 2026 is $81,500, up 7% from last year per NACE projections. That's a fine raise. It's also not $300,000.
What you're actually looking at is 2 separate markets that share a job title. The top of the market, engineers who can build AI infrastructure, tune models, or do forward-deployed work at clients, is on fire. LinkedIn data from April 4 shows forward-deployed engineer roles hit 8,500 new positions, a 42-fold increase since 2023. The rest of the market is recovering, but it's recovering into a much more crowded room. Taylor said it himself: "Way more people have pursued computer science. The jobs haven't disappeared, but competition for them is dramatically higher than it was even five years ago."
Marc Andreessen posted last week that AI job loss fears are "all fake" and that AI will drive job growth. I'll grant him the first part: the data does not support mass displacement right now. But calling the fears fake is doing a lot of work to avoid the harder question, which is whether this hiring surge is durable or whether it's a capital-allocation spike that reverses when AI investment cycles cool.
What the Data Doesn't Tell You
The 67,000 open roles are concentrated in firms where AI investment is currently flowing. That's a real signal, not noise. But it's also a cohort of companies that are spending aggressively on the assumption that AI infrastructure built today pays off in 3 to 5 years. Some of those bets will hit. A lot of them won't. Startup failure rates haven't changed because the hiring market got hot.
Taylor's own hedge is the honest one: "Maybe AI compresses some roles entirely. Or maybe it makes great engineers so leveraged that companies fight even harder over them." He doesn't know. Neither does Andreessen. Neither do I.
Here's what I'd actually tell a builder trying to make a career decision right now: the broad recovery is real enough to stop catastrophizing, but not broad enough to coast on a generic backend skillset. The salary data points to one clear conclusion. Companies are paying a premium for engineers who understand how AI systems fail in production, not just how to call an API. That's the gap worth closing.
The boom is real. It just isn't for everyone yet, and the window to get on the right side of it is shorter than the headlines suggest.
