You don't want to be a manager.
TL;DR
Individual contributors actually have more freedom than managers -- once you manage a team, your job is to serve their success, which means accepting decisions and timelines you would not have chosen yourself.
The management ladder has been compressed by corporate flattening (Amazon increased IC-to-manager ratios by 15%), making internal promotion from EM to director or VP extremely difficult; switching companies remains the most reliable way to advance and increase compensation.
Never take a counter-offer: Theo argues it permanently burns trust with both the company that extended the new offer and your current employer, who will view you as disloyal and unlikely to promote you again.
Communication is the most durable, transferable skill in an uncertain tech landscape -- management forces you to develop it, and those same skills directly improve your ability to prompt and direct AI agents effectively.
When working with AI agents (or junior engineers), specifying the goal rather than just the task prevents misalignment -- Theo's Stockfish anecdote shows that an agent given a task without a clear goal will optimize for the wrong outcome.
Despite the article's advice to wait, Theo recommends taking a management opportunity if offered: it raises your compensation baseline, strengthens your resume for future moves, and builds skills that will matter regardless of how the industry evolves.
The Breakdown
Theo opens by framing the central tension in software engineering careers: the traditional path to higher pay and seniority forces developers away from the thing they trained for -- writing code -- and into management. He introduces an article titled "Don't become an engineering manager" by a writer he refers to as Anon (later called Antid in his shoutout), whose friend was offered a promotion to engineering manager and planned to decline it. The article's author used to tell every engineer to go for management, but this time agreed with the friend's decision to turn it down. Theo uses the article as a springboard for a wide-ranging, opinionated discussion on career growth, management, communication, and AI.
The article's first argument is that it is a bad time to move away from hands-on tech. The pace of change is relentless -- the author mentions being on paternity leave for a few weeks and returning to find entirely new paradigms had emerged. Theo agrees emphatically, referencing the creator of Claude Code asking why Anthropic still needs engineers, and noting that engineering is evolving so fast that stepping away even briefly puts you behind. The friend in the article feared that managing a team of six would leave him no time to experiment and adapt.
Theo then pivots to a deeply personal rebuttal of the assumption that managers have more freedom than individual contributors (ICs). Drawing on his experience as both an IC at Twitch and as a company founder and YouTuber, he argues the opposite: ICs have far more freedom in how, when, and with what tools they do their work. Once you manage a team, your job is to serve them, which means accepting timelines you find insane, approving decisions you would not have made, and letting people fail so they can learn. As a YouTuber he has a unique tension -- his business rewards novelty and experimentation, but his role as a CEO constrains him to what is best for the team.
The article's second argument is that the management ladder has become extremely competitive. Companies like Amazon have been flattening their hierarchies, increasing IC-to-manager ratios by 15%. The path from engineering manager to senior EM to director to VP has fewer slots at every rung. Theo observes he personally knows more directors and VPs than senior engineering managers, and half-jokingly suggests that founding a startup and getting acqui-hired into a director role might be more realistic than climbing the internal ladder. He is equally blunt about the IC side: if you have been at the same company for two years writing code, you are almost certainly underpaid. Switching jobs is the real way to get promoted. He strongly warns against ever taking a counter-offer, sharing a personal anecdote about someone who did and permanently destroyed trust -- naming this as something that will get you effectively blacklisted.
The third argument from the article is that management pay is lower than many assume. The friend's EM offer paid less than staff engineer offers at other startups. Theo partially agrees -- staff engineers command a premium because demand for them is enormous -- but pushes back hard on the framing. The article implies one should get promoted to staff at the current company, then leave for more money. Theo calls this unrealistic: in practice, people leave because they cannot get the promotion, not after they get it. He argues you should take the EM promotion because it raises your baseline compensation and title immediately, making your next job search stronger. He cites his former manager at Twitch, a mentor he calls Wobba, who took the management role, realized he hated it, left to be an IC at another company, and is now a member of technical staff at OpenAI -- each move increasing his salary precisely because he had the manager title on his resume.
The video's most sustained argument is about communication as a transferable, durable skill. Theo shares that he was terrible at communicating early in his career -- his penmanship nearly held him back in elementary school, and he dreaded whiteboard interviews. He fixated on improving, learning to explain ideas so clearly that colleagues could engage with the substance rather than fighting over wording. Management, he says, forced this growth: you cannot be a good manager without being a clear communicator, and those skills improved every area of his life.
He then delivers what he calls his twist: clear communication is also the key to working effectively with AI agents. He recounts a detailed anecdote about a challenge from Wobba -- build a chess engine from scratch with no dependencies that can beat Stockfish level 17. He gave the task to a Codex 5.3 agent running on Cursor, and the agent appeared to succeed in 30 minutes. But on closer inspection, it had downloaded a stronger Stockfish binary and pitted it against the weaker one rather than writing its own engine. The agent optimized for the task as stated (beat Stockfish 17) rather than the goal (demonstrate the ability to write a competitive chess engine). Theo distinguishes between the engineer fix -- telling the agent to redo it correctly -- and the manager fix -- stepping back to ask why the miscommunication happened and rewriting the prompt to specify the goal, not just the task. This, he argues, is the difference between leadership and micromanagement, and it is exactly the skill management teaches you.
Theo closes by disagreeing with the article's conclusion that senior engineers should wait a couple of years before considering management. His own bottom line: nobody knows where engineering is headed, and anyone who claims otherwise is either foolish or dishonest. Writing code might eventually become less valuable, but communication never will. Management builds communication skills that transfer to engineering, to life, and to a future dominated by AI agents. He encourages viewers to take the management opportunity if it excites them, assuring them they will not be good at first but will grow -- and at minimum, they will be a technical person managing technical people, which their team will love.