Will Nvidia’s $100B Bet on OpenAI Leave Tesla Behind?
Overall Thesis
Nvidia's $100B OpenAI deal is a financially brilliant move that adds ~$1.25 trillion in market cap for $100B investment, but it raises concerns about Tesla and xAI's chip supply; AMD remains a serious underestimated competitor and AI itself may shrink Nvidia's CUDA moat.
Narratives
Nvidia's $100B OpenAI investment is a financial no-brainer adding ~$1.25T market cap at current 50x PE, but its CUDA moat may shrink as AI makes software porting easier, leaving room for competitors.
Key Arguments
- 5M GPUs at ~$35k margin each equals ~$150B margin over 3 years.
- At 50x PE, $25B annual EBIT addition creates ~$1.25T market cap.
- Nvidia added $100B in market cap immediately on announcement.
- CUDA moat could erode because AI makes software translation easier.
- Nvidia is a tremendous company that will not give up its lead.
Tesla and xAI may face chip supply concerns because of Nvidia's OpenAI deal, though Elon believes he will still be first to 10, 100, and 1000 gigawatts of coherent training compute.
Key Arguments
- Elon initially said the deal is concerning then claimed xAI will still lead in compute scaling.
- Elon dropped Dojo because Nvidia is so much better.
- Tesla/xAI still need Nvidia chips and supply could be constrained.
AMD is underestimated and making good progress on inference; CUDA's advantage is shrinking because AI makes software translation easier.
Key Arguments
- AMD is very aggressive and not dummies.
- Making good progress on inference workloads.
- Highly more competent than Tesla team as a giant semiconductor company.
- AI shrinks CUDA moat advantage.
OpenAI gets $100B investment from Nvidia, a no-brainer deal for them.
Key Arguments
- OpenAI is one of few companies willing and able to do this scale of deal.
- Gets $100B to buy Nvidia chips.
xAI may be disadvantaged by OpenAI getting Nvidia funding but Elon claims xAI will be first to 10 and 100 and 1000 gigawatts of compute.
Key Arguments
- Elon says xAI will be first to 10 gigawatts, 100 gigawatts, 1 terawatt.
- Chip supply concerns given Nvidia favoring OpenAI.