ExpertsFM#403 How Jensen Works
Founders · October 20, 2025
5 Selected Lessons:
heuristic · 2 reasoning_pattern · 1 framework · 1 case · 1
Top five emails — unfiltered signal from the organizational edge
Source
Founders — #403 How Jensen Works (October 20, 2025) · at 27:07 ▸
When this applies
when a leader of a large organization needs ground-truth information that isn't filtered or distorted by management layers
The takeaway
Create a low-friction, structured channel for raw information from the frontline, and read it personally rather than relying on summarized reports.
Why it works
Direct, lightweight reporting from every level bypasses the incentive of middle managers to filter bad news, surfacing weak signals (like early GPU-for-ML use cases) before they become obvious.
One level up
Strategy should be inferred from what the organization is actually doing at the edges, not dictated top-down and assumed to be followed.
Where it stops holding
This stops working when volume exceeds what a leader can personally process, or when employees self-censor because they fear the information will be used against them.
Detail
procedure:
- Ask every employee to send top 5 things they're working on, with observed market signals (pain points, competitor activity, tech developments, delays)
- Format as five action-word bullet points (finalize, build, secure)
- Tag emails by topic in the subject line for searchability
- Read ~100 per day personally to extract strong signals
rationale: Strategy isn't what I say, it's what they do — he wants information from the edge, not filtered through layers of management.
In their words
he wants is information from the edge. That is a direct quote from him. He wants information from the edge.
Tortured into greatness — self-criticism and pain-tolerance as character-builder
Source
Founders — #403 How Jensen Works (October 20, 2025) · at 17:10 ▸
When this applies
when deciding how to respond psychologically to both success and failure to build durable organizational character
The takeaway
Treat the capacity to endure pain and setbacks as a core skill to cultivate deliberately, valuing it above raw intelligence.
Why it works
Habitual self-criticism prevents complacency from taking root even after wins, and enduring repeated pain/setbacks builds resilience that correlates with eventual success.
One level up
Character, not talent, is the scarce resource that determines long-run outcomes.
Where it stops holding
This stops working when self-criticism becomes corrosive rather than motivating, destroying morale or confidence instead of building resilience.
Detail
approach: Deliberately maintain low expectations and treat pain/suffering as valuable training for resilience, applying the same scrutiny to oneself as to others.
signals:
- Low expectations correlating with high resilience
- Treating setbacks as opportunities
- Standing up after a great quarter and still saying 'you suck'
rationale: Greatness comes from character, and character is not formed out of smart people. Character is formed out of people who suffered.
In their words
I think one of the great advantages is I have very low expectations. Most of the Stanford graduates have very high expectations. People with high expectations have very low resilience. And resilience matters in success.
Teach constantly (Professor Jensen / 90% teaching rule)
Source
Founders — #403 How Jensen Works (October 20, 2025) · at 0:00 ▸
When this applies
when leading any organization where alignment and speed depend on shared understanding
The takeaway
A leader's primary output is clarity transmitted to others, not decisions made alone.
Why it works
Constant teaching replaces top-down directives with shared mental models, so employees can act independently yet consistently with strategy.
One level up
Leverage scales when understanding, not just instructions, is distributed.
Where it stops holding
This stops working when the leader's own understanding is unclear or inconsistent, since teaching then spreads confusion rather than clarity.
Detail
rule: Spend the vast majority of leadership time teaching the organization, repeatedly and pervasively, so employees internalize strategy.
threshold: Jim Sinegal: if you're not spending 90% of your time teaching, you're not doing your job.
rationale: Employees who understand strategy directly can act fast without waiting for instructions.
In their words
He says, if you're not spending 90% of your time teaching, you're not doing your job.
Criticize publicly so the whole organization learns from one mistake
Source
Founders — #403 How Jensen Works (October 20, 2025) · at 13:21 ▸
When this applies
when deciding whether to address a mistake privately or use it as a teaching moment for the whole organization
The takeaway
Optimize feedback for organizational learning, not for individual comfort.
Why it works
Public criticism converts one person's error into a shared lesson for everyone, preventing the same mistake from recurring elsewhere and discouraging political turf-protection.
One level up
Treat the work, not people's feelings about the work, as the thing to be protected.
Where it stops holding
This stops working when public criticism is perceived as humiliation rather than teaching, destroying trust and psychological safety instead of building shared learning.
Detail
rule: Criticize mistakes publicly at company meetings rather than privately, so the entire organization learns.
rationale: We are not optimizing for not embarrassing somebody. We're optimizing for the company learning from our mistakes.
In their words
We are not optimizing for not embarrassing somebody. We're optimizing for the company learning from our mistakes.
Swarm your greatest opportunity — CUDA's decade-long, expensive bet
Source
Founders — #403 How Jensen Works (October 20, 2025) · at 45:26 ▸ · at 50:16 ▸
When this applies
when a leader spots a weak signal of a potentially enormous future opportunity that requires years of expensive, unproven investment before payoff
The takeaway
When you identify a weak signal pointing to a potentially vast future market, commit heavily and for a long duration, even against investor pressure and margin damage, rather than waiting for certainty.
Why it works
Early, sustained, expensive investment in a capability (CUDA) that lowers the barrier to adoption compounds over time, because expanding ease-of-use expands the addressable market faster than competitors can respond.
One level up
Long-duration conviction-based bets, sustained through adversity, are what create category-defining moats — not single decisive moments.
Where it stops holding
This stops working when the underlying signal doesn't actually represent a real emerging market, in which case sustained heavy investment simply destroys capital with no payoff.
Detail
situation: Researchers began hacking GPUs for non-graphics computation in 2002; the market for this was unproven and tiny.
decision: Jensen hired the researcher who coined GPGPU, built the CUDA programming model to make GPUs easier to program for general use, and pushed the investment hard even as gross margin fell from 45% to 35% during the 2008 financial crisis, against investor demands for a course correction.
outcome: CUDA became the foundational technology enabling NVIDIA's dominance in AI/GPU computing, eventually creating trillions in value.
lesson: Sustained, expensive, long-duration investment in a weak-signal opportunity, maintained through external pressure to abandon it, can create a category-defining moat decades later.
In their words
Jensen believed so strongly in the market potential that he remained committed to the course he had chosen, even as his investors demanded a strategic course correction.