The Nvidia Way, Tae Kim
- Substrat: flat organisation with very high standards
- strategy: build the future of computing
- tactics:
- anticipate the next innovation cycle
- 6 months iteration cycles (3 parallel teams)
- very high standards (long hours, speed of light timelines, paranoid)
- pilots in command (pic, project leads that report to CEO)
- free product handouts to hardware makers
- flat hierarchy (CEO has >60 directs) with nimble, continuous planning
- "Top Five E-Mail": weekly employee email with five bullet points where the first word is an action word (e.g. finalise, build, secure); each department to add its designation in the subject line; Jensen reads >100 Top-5-emails per day
Notes:
- competitive advantages
- backwards compatibility: early mistake to expect developers to switch from inferior standard to unknown proprietary one that required switching costs
- fast iteration cycles
- parallelised chip development: 3 teams in parallel (1 for new chip, others develop derivatives)
- later added Nvidia Research to add moonshot development in parallel
- hardware emulation allowed emulating hardware features using software
- very high standards, e.g. on work hours
- told new employees that they must be "ultra-aggressive", constant push push push
- speed of light: ideal timeline as dictated by the laws of physics (without delays, etc.)
- at each monthly company meeting, he said "We're thirty days from going out of business"
- flat hierarchy with continuous planning: he has >60 directs, he rarely does 1-on-1s, he criticises individuals in company-wide meetings, "We just plan continuously. There's no five-year plan."
- Pilot in Command (PIC): strategic project lead that reports directly to Jensen
- ship the whole cow: create derivative low-cost products from rejected parts
- free product handouts: handed out free chips to Intel, Microsoft, monitor companies
- teach the market: CUDA also on low-cost GPUs (which was expensive) & courses at universities
- Top Five E-Mail: weekly employee email with five bullet points where the first word is an action word (e.g. finalise, build, secure); each department to add its designation in the subject line; Jensen reads >100 Top-5-emails per day
- new employee receives a brokerage account to get restricted stock units (RSUs)
- end of year 1: employee vests 25% of initial grant; after that, 25% of annual grant every quarter
- annual refresher grant (vests over 4 years) if an employee receives an "outstanding" rating
- "top contributor" (TC) rating: employee receives special grant with Jensen in CC (vests over 4 years)
- Jensen may award grants unilaterally at any time
- Jensen was around 30 at time of founding Nvidia
- questions from Don Valentine to founding team: what is Nvidia? what will Nvidia be in 10 years?
- ideal product presentation: easy to understand, clear alternative, evokes emotions
- the company never discounted its chips unless it got something in return
- quotes
- "I can tolerate a lot of discomfort."
- "I have a very long-term horizon. I can be impatient about certain things, but infinitely patient about others."
- "People with very high expectations have very low resilience. Unfortunately, resilience matters in success."
- "I wish upon you ample doeses of pain and suffering."
- "There may be people smarter than me, but no one is ever going to work harder than me."
- "Make no mistake. Intel is out to get us and put us out of business. (...) We need to go kill Intel." (at the time, Intel hat 860x Nvidia's revenue)
- "Second place is the first loser."
- "Just go win."
- "I don't like giving up on people. I'd rather torture them into greatness."
- "I look in the mirror every morning and say 'you suck'."
- "You have to allow yourself to be obsessed with your work."
- against opposition (2013): "Deep learning is going to be really big. We should go all in on it."