I came to NVIDIA looking for a crack in the strategy. I found pressure points. I did not find the simple mistake I had hoped would explain the whole story.
What changed was not just my opinion of a company. Its example began to interfere with my own idea of what I was allowed to attempt.
Not my favourite band
As a teenager, I liked other bands more than Nirvana, the Sex Pistols or The Prodigy. Yet Cobain, Sid Vicious and Keith Flint left something that outlasted the playlist: a different sense of how a person could think, act and refuse a convention.
That is my comparison with Huang. Not the biography. Not a verdict on musicianship. Not the jacket. Influence does not require first place in your personal favourites.
What versus how
For me, the familiar Jobs analogy concerns the finished object: here is a thing I want. Huang produces a different reaction: here is a possibility that makes my previous question feel too small.
This is a description of my response, not an objective ranking of two careers. I do not have to understand every engineering decision for the change in scale to disturb my own assumptions.
to own?
to attempt?
The rebel was already inside
Christensen gave the book its starting question: can the habits that make a company exceptional become constraints on its next transformation?
But Huang is not merely an example of the organisation to which that question applies. In Chapters 13 and 15, he is part of the resistance to its easiest answer. The person I might have advised to find an internal rebel was already playing that role.
That does not make every choice right. It makes the analysis more demanding.
Read Chapter 13 →A method, not a costume
What I want to take from the example is not a keynote persona. It is a willingness to expose the reasoning behind a decision, to test an unfamiliar market, and not to treat the current model of success as the only available model.
Huang’s 2024 Stanford conversation is useful reading here: he distinguishes early signals of a new undertaking from the financial results that follow. My book’s margin-centred account has to engage with that public description, rather than assume it away.
Stanford GSB conversation · 25 April 2024 ↗
I am not borrowing permanent exhaustion as an ideal. A useful disturbance is one that results in work, not one that leaves you too distracted to finish any.
Admiration without surrender
A platform can expand the space in which I can act and create dependencies I should examine. Both can be true.
I do not need to dislike NVIDIA in order to ask about interoperability, access, price or control. Nor does recognising an achievement oblige me to accept its next decision without inspection.
The point of a sparring partner is not to let you win. It is to make a weak explanation difficult to keep.
Back to the desk
My reaction tends to arrive in two stages. First: how did my imagination fall behind again? Then: which question can I actually test now?
For me, that question came from banking. An agent may be able to perform a transaction without holding the authority to perform it. What changes when the delegation is revoked while the agent is still running?
The narrower experiment became AuthorityClaw. The afterword follows that path, including the post-challenge checks that found problems the demonstration had not covered.
From book to experiment →Take the method, leave the merch
I do not need Huang’s answers to become my answers. I want the example to make my questions less timid — and the results less easy for me to excuse.
The useful form of admiration ends neither at the stage nor at the checkout. It sends me back to a problem I understand well enough to make specific.
Turn the disturbance into a question.
Turn one question into a test.
Finish the test.
That is why the music analogy matters to me. It concerns what remains after the performance: a change in how I act.