The release of Anthropic's Claude Code and Opus model in late 2023 is considered a second "ChatGPT moment" that sparked massive adoption among software developers.

According to Edward Chancellor's capital cycle theory, the telephone industry is the only historical example of a technology boom where the infrastructure builders did not ultimately go bust, due to i...

Kai Wu posits that the true moat for major software companies has always been distribution and customer lock-in, not their underlying code.

Kai Wu argues that the prevailing market narrative is that the more powerful AI becomes, the more traditional software companies will be disrupted.

The CapEx-to-sales ratios for many hyperscaler companies are currently higher than AT&T's was at the peak of the dot-com boom.

Kai Wu believes the LLM layer of the AI stack is at risk of commoditization due to the rapidly closing performance gap between competing models.

OpenAI is partnering with private equity firms to deploy engineers into enterprise companies to accelerate the adoption of its products.

Kai Wu asserts that AI model providers like Anthropic and OpenAI are subsidizing token costs, selling them below their actual cost in an effort to capture market share.

NVIDIA's high profit margins are attracting significant competition from companies like Cerebras and NVIDIA's own customers, including Google and Amazon, who are developing their own chips.

Kai Wu's base case is that the Magnificent Seven companies will undergo a valuation re-rating to lower multiples as they transition from asset-light to capital-intensive businesses.

The Fama-French value factor, which produced excess returns for approximately 80 years, ceased to be effective after its publication in 1994.

Kai Wu suggests that current enterprise AI adoption is driven by a "fear of missing out" (FOMO) and pressure from boards, and its sustainability depends on generating demonstrable ROI.