From: allin
Nvidia has experienced an “incredible run,” with its recent earnings showing significant growth across the board [08:45:00]. This performance has led to unprecedented revenue scales in the history of business [08:30:00].
Unprecedented Financial Performance
Nvidia’s revenue reached 16.6 billion and a gross margin of 75% [08:20:00].
Despite these strong results, there are indications that quarter-over-quarter revenue growth has started to cool off after a period of massive boom [08:37:00].
Sustainability of Growth
The sustainability of Nvidia’s growth is a key discussion point. Reed Hoffman, a venture capitalist and board member at Microsoft, believes the current level of “pure heat” in growth is sustainable for about two years, which in public market terms is considered “forever” [09:10:00].
Key factors influencing sustainability include:
- Chip Dominance: Nvidia holds a “very sharp lead” in chips crucial for AI training clusters [09:25:00].
- Inference Chips: While effective in inference, the bulk of future demand will likely be for inference chips, where more competitors are expected to emerge [09:33:00]. This will challenge Nvidia to either maintain prices and margins or respond to a more competitive market [09:54:59]. This competitive pressure is expected to start playing out within one to two years [10:07:00].
- AI Infrastructure Investment Rationale: Large cloud computing platforms, such as Microsoft, approach AI infrastructure buildout with a blend of strategic insight and sensible return on capital [12:08:00]. The investment is driven by the understanding that AI represents a platform change for productivity and cloud services, necessitating participation [12:41:00]. However, there’s also a focus on rationalizing capital and generating revenue sooner, rather than “spending like drunken sailors” [12:47:00].
Future Outlook and Competition
Experts advise against shorting Nvidia due to its strong current position and continued growth potential, even as competition is anticipated [10:18:00]. The market is not expected to be dominated by a single “God model” [17:51:00]. Instead, a “network of models” approach is likely to prevail, where smaller, specialized models are trained using high-quality data, often with assistance from larger models, to achieve specific, cost-effective outcomes [18:59:00]. This dynamic suggests ample room for various startups and large companies to succeed, not just a few dominant players [17:11:00].