Action Playbook

Draw Your Flywheel on One Slide

Huang has presented the same flywheel for 20 years. "100% of Nvidia's strategies." Can you describe yours on one slide? If it doesn't loop, you don't have a flywheel.

Source: CUDA Flywheel

Reframe Price as Cost

Stop defending pricing. Calculate TCO. A $50B factory 10x more efficient is cheaper than a $30B one at 1x. Make CFOs your audience.

Source: Cost vs Price Framework

Implement Speed of Light for Key Projects

Pick your three most important initiatives. Calculate the theoretical minimum time to complete each. Make the gap between actual and theoretical visible. Hold the team accountable to closing it.

Source: Speed of Light Culture

Replace Status Reports with Raw Signal

Implement T5T or equivalent. Encourage all levels to share priorities directly. Read a sample daily. Look for weak signals. The problems people mention in passing are the ones that matter most.

Source: Top 5 Things System

Identify Your CUDA: The Long Bet

What investment would create compounding ecosystem advantage over 5–10 years? The one Wall Street would hate but physics supports? Make that bet.

Source: CUDA Decision (2006)

Build Full-Stack or Partner Full-Stack

Chips alone don't win. Hardware + software + tools + integration + forward engineers = moat. If you can't build every layer yourself, partner strategically to complete the stack.

Source: Full-Stack Philosophy

Prepare to Cannibalize Yourself

Identify the technology that could disrupt your current offering. Build it or acquire it. Control the transition rather than defending the legacy. The company that eats itself survives; the one that protects itself gets eaten.

Source: Groq Acquisition

Make Intellectual Honesty Structural

Build flat hierarchies, group meetings, direct channels. Reward early problem-surfacing. The Sega call saved NVIDIA because someone was honest about a failing architecture before it was too late.

Source: Intellectual Honesty Culture