Software Engineer Jobs Are Rising, Not Dying — Box CEO's Explanation

AI influencers who've never shipped a product keep saying AI is killing software engineering. The job market says the opposite — engineering job posts are on the rise. Aaron Levie (CEO of Box) has the cleanest explanation of why: vibe coding expands the total surface of software that needs to exist.

The Levie Mechanism

The argument in one paragraph:

AI vibe coding drops the cost of creating software
That democratizes creation — engineers ship more, non-technical teams prototype their own tools
Non-technical builders hit the wall: "there are still 50 other tasks that go into building and maintaining a product that works"
They hire engineers to take over the vibe-coded projects
All this work wouldn't have existed without vibe coding in the first place

That last point is the one most people miss. The projects engineers are inheriting aren't cannibalizing existing work — they're net-new demand that only exists because non-technical people could start them.

Why AI Didn't Shrink the Engineering Market

Two forces push in opposite directions:

Force	Effect
AI makes engineers more productive	Could shrink demand (same output, fewer engineers)
AI makes non-engineers start more projects	Expands demand (more projects need engineers)

Levie's argument is that the second force is dominating — because most non-technical vibe-coded projects eventually need real engineers to reach production. The things that show up in production software are:

Proper auth and session management
Multi-tenant data isolation
Error handling and retries
Observability
Deployment pipelines
Security review
Performance tuning
Backwards compatibility
Customer support for edge cases

Vibe coding helps with step 1 of 50. The other 49 still need engineers.

The Hidden Expansion

The deeper shift: vibe coding is expanding the total surface of software that exists. Projects that would never have been staffed ("let's build a tool to track this internal workflow") now get started because the activation energy is near zero. Some percentage of those get real traction, outgrow the vibe-coded prototype, and then need a real engineer to productionize them.

This is the pattern Salesforce/HubSpot hit with no-code: the market didn't shrink for developers, it grew — because more companies started automating more processes once the floor dropped.

Who's Wrong Here

The take "AI is killing software engineering" tends to come from:

AI influencers who aren't engineers and haven't had to ship a product that stays up
VCs selling a thesis that AI kills labor costs
Bench-marking the first 5% of the work (generation) and ignoring the other 95% (maintenance, integration, scale)

The people actually running engineering orgs — the Levies, the Eran Zinmans of the world — are hiring. The data aligns with them.

Key Takeaways

Software engineer job posts are rising, not falling — the market is expanding
Aaron Levie's model: vibe coding democratizes creation, which generates more projects, which eventually need real engineers to productionize
The hidden mechanic is net-new demand: projects that would never have been started at all now get started
Generation is ~5% of shipping software; the other 95% (maintenance, reliability, security, scale) still needs engineers
Listen to operators, not influencers, on labor market questions

Resources

Aaron Levie on X — Box CEO, source of the "50 other tasks" framing
Box — Levie's company
State of Software Jobs trend data — Industry tracking for cross-reference on job market claims