Keep pulling the thread on Ev.
AI inference platforms are securing 9-figure contracts with startups, a scale rarely seen in traditional SaaS.
In the current AI market, high gross margins in an AI product are viewed negatively, as it implies that customers are not using the costly inference-based features.
The traditional "spreadsheet investing" rules for software, such as high gross margins and capital efficiency, are no longer applicable in the AI era.
In the current AI paradigm, companies can achieve revenues well over $1 billion without having proven unit economics or durable product differentiation.
The rapid revenue growth of new AI companies, with some scaling from 1 to 30 to 300 million in ARR, is primarily enabled by inference-based business models.
Ev believes AI agents represent the most significant product and business model innovation since the beginning of SaaS.
The release of Anthropic's Opus 4.5 model last winter was a genuine breakthrough that led to parabolic revenue growth for the company and its Claude Code product.
If AI model capabilities hit a ceiling and open-source models achieve 95% of that performance, it would be a very threatening scenario for frontier labs like OpenAI and Anthropic.
Starlink's consumer and B2B broadband business now constitutes the vast majority of SpaceX's business, surpassing its original launch services.
A potential liquidity event for Anthropic at a $1.5 trillion valuation would generate a gross return 35 times larger than the successful Snowflake pre-IPO round.
Some individual investors have invested as much as $3 to $4 billion into Anthropic.
Crusoe's business model involves constructing its own data centers by acquiring power, land, and permits, often for hyperscaler customers.