Strengths & Weaknesses

CUDA was a billion-dollar bet with zero paying customers. Most CEOs would kill it. Jensen held the line for years until the market materialized. That single decision created a $5T company.

No competitor offers chips + interconnect + networking + software + ecosystem. Intel tried. AMD tried. Google TPUs are narrow. NVIDIA is the only company that owns the entire accelerated computing stack from silicon to frameworks.

2+ hour keynotes without notes, arena audiences. Can explain chip architecture to investors and sovereign AI to heads of state. One of the few technical founders who can hold an arena audience for two hours without notes.

55+ direct reports, reads 100 T5T emails daily, no status reports, no 1:1s. Startup founder intensity at multi-trillion scale. Most companies ossify into bureaucracy at a fraction of this size. NVIDIA has not.

Hundreds of thousands of CUDA projects, integrated into every cloud provider and AI framework. A self-reinforcing flywheel where every new developer makes the platform more valuable for every other developer.

Acquired Groq ($20B) to add non-GPU inference chips. Expands total addressable market rather than protecting existing products. Jensen would rather disrupt himself than let a competitor do it.

"My heart rate goes down when the world is falling apart." From reform school to near-bankruptcy multiple times. The company nearly died building the Riva 128 and again with the Xbox fiasco. Each crisis forged deeper conviction.

Four hyperscalers account for a disproportionate share of 90% data center revenue. When your biggest customers are also your most motivated competitors, the relationship is structurally fragile.

Every chip manufactured at TSMC. 70%+ of CoWoS-L advanced packaging capacity. No plan B for Taiwan disruption. A geopolitical risk that no amount of engineering excellence can mitigate.

"Even when free, not cheap enough" is dismissive. Google TPUs handle 75% of Gemini inference. Amazon Trainium offers 30–40% better price/performance. The moat is real but not infinite.

Lost 45% China market share. $5.5B in write-downs. Public criticism of export controls is politically risky in an environment where both parties favor hawkish China policy.

"Speed of light" culture and flat structure with 55+ direct reports creates high intensity. What works for Jensen (who thrives on pressure) may not be sustainable for the broader organization over decades.

Leather jacket, arena keynotes, rock-star treatment make it harder to separate signal from spectacle. When the messenger becomes the message, critical analysis of strategy gets lost in the charisma.

At $4.4T+, priced for AI spending growing faster than it already is. Any plateau, even a deceleration in the rate of growth, becomes existential for the stock, regardless of underlying business health.