🤔

Continuing from a recent post about how Oracle forced engineers to train AI and then laid off 30,000 people to save on bonuses — if you thought this was a one-off act of corporate greed, welcome to the structural restructuring of the market.

Brian Armstrong (CEO of Coinbase) published a letter casually announcing the layoff of 14% of staff. They're not just cutting costs. They're restructuring the company for an AI-native format. Key takeaways from his letter:

1️⃣ The death of "pure managers". Zero "pure" managers. Now every lead at Coinbase must be an individual contributor. If for the last three years all you did was organize daily standups, move tickets, and ask "how are you" in 1-to-1s — your position is eliminated. A manager is now a player-coach up to their elbows in code.
They've kept a maximum of 5 hierarchy levels from CEO to bottom. Layers slow down work and create a "coordination tax."
2️⃣ One-person team. The company is transitioning to micro-teams and "one-person teams" (engineer + AI agents replacing design and product management).

And this isn't just Armstrong's local quirk. It's a trend. Vedomosti published an excellent market overview. In February, Apple acquired the startup invrs.io, which consisted of exactly one employee. Corporations are now buying not teams, but the expertise of one person multiplied by AI.

At YCombinator, 25% of startups in the current batch are 95% composed of generated code. Solo founders break even within six months. A "working product" is no longer worth anything — it can be assembled over a weekend at the cost of a couple cups of coffee. Against this backdrop, the combined valuation of platforms for this (Cursor, Lovable, Replit) soared past $36 billion last year alone.

Businesses have realized they no longer need huge galleys of developers and armies of scrum masters to somehow row those galleys in one direction. They need one person with product thinking who, using AI tools, can cover the tasks of an entire department from five years ago.

So programmers are almost no longer needed? 😲
Not so fast.

In the same article, there's a great figure from Veracode: AI creates 9 times more vulnerabilities than a human developer. Leaked keys, leaky authentication, childish architectural errors. AI doesn't understand security context; it just happily executes commands.

So here's the picture: businesses cut costs, throw out parasitic managers and mid-level coders. Soloists churn out MVPs made of duct tape and sticks at cosmic speeds. And it will be up to the remaining hardcore engineers to clean up the mess.

Well, at least there will be fewer stupid meetings and approvals. That's a plus. 🤷‍♂️