← All episodes

6 Levels of AI Investing Everyone Should Know

| 12 products mentioned
Watch on YouTube ai investing venture capital large language models ai infrastructure business strategy competitive moats commoditization

Lonsdale, Moshkovich, and Kolicich break down AI investing through a six-level framework spanning from energy infrastructure to end-user applications, arguing that while LLM companies command massive valuations, many are becoming commoditized in certain domains. The hosts challenge the conventional wisdom about AI model valuations by examining how different companies (OpenAI, Anthropic, Google, Meta, and X) are positioning themselves around distinct data assets and business models rather than competing purely on model capability.

Key takeaways
  • The six-level AI stack progresses from energy → chips → data centers → LLMs → software infrastructure → applications, with the most defensible value often concentrated at Level 5 (applications) rather than in the model companies themselves.
  • AI models exhibit commodity-like characteristics in certain use cases (document processing, structured output) where open-source alternatives work well, while remaining premium in frontier domains like coding and reasoning where proprietary models command pricing power.
  • Current LLM company margins (50-60% on APIs) far exceed true commodity businesses (5-15%), but the critical question is whether this durability persists as capabilities commoditize—a timeline dependent on how quickly rivals catch up.
  • Model companies' defensibility will depend less on raw capability and more on building product moats and data gravity (memory systems that lock in customers through personalization), similar to how AWS moved beyond commoditized compute primitives into managed services.
  • Different AI labs pursue divergent business models: OpenAI focuses on consumer, Anthropic on developer infrastructure, Google on productivity suite integration, Meta on content/culture, and X/Grok on reasoning and physical-world robotics—making them less direct competitors than ecosystem players.
  • Investor incentives heavily bias narrative around AI valuations—those with billions deployed want models to remain scarce and valuable, while infrastructure players benefit from positioning them as commodities, creating systematic bias in public discourse.

Mentioned (12)

Nvidia
Nvidia "Level one is chips. Obviously Nvidia dominant but there's been a lot of new things there we're ex..." ▶ 0:17
Anthropic
Anthropic "Level three is where we put on top of those is the actual LLM companies. is actually anthropic an..." ▶ 0:29
Gemini
Gemini "Level three is where we put on top of those is the actual LLM companies. is actually anthropic an..." ▶ 0:29
OpenAI
OpenAI "Level three is where we put on top of those is the actual LLM companies. is actually anthropic an..." ▶ 0:29
xAI
xAI "Level three is where we put on top of those is the actual LLM companies. is actually anthropic an..." ▶ 0:32
EC2
EC2 "So if you look at the hyperscala business, the primitives of it, EC2 for compute and S3 for storage." ▶ 3:36
S3
S3 "So if you look at the hyperscala business, the primitives of it, EC2 for compute and S3 for storage." ▶ 3:36
Amazon Web Services
Amazon Web Services "Uh so if you go to Amazon web services, there's two primitives. S3 is like a storage primitive." ▶ 3:43
G Suite
G Suite "And if you are a G Suite user or if you're a Gmail user, chances are these products are going to ..." ▶ 8:25
Grok
Grok "So I like to say that X is in the truth business and Grok today is very much trying to position i..." ▶ 9:13
SpaceX "They also think they're the best at building infrastructure. And so it's like not only is Elon bu..." ▶ 9:50
Tesla
Tesla "Yeah, exactly. I shouldn't leave out Tesla. You have these amazing robots coming." ▶ 10:12