digestVidSign in
summary

Fox Business - These models are getting ‘CLOSER’ to ‘training themselves’: Spear Invest founder

Thumbnail of the video “Fox Business - These models are getting ‘CLOSER’ to ‘training themselves’: Spear Invest founder”
Community upload. AI output has not been independently verified.
Author: Fox Business
Categories: Business Technology Personal Finance
Tags: business, technology, finance
Views: 2 · Last viewed: 2026-10-16 13:32 UTC

Short Summary

Main recommendation: Position in AI-related infrastructure now, especially optical networking names such as Coherent, Lumentum, and Credo, while treating Nvidia as a strong risk-reward play that benefits from both open and closed AI models. The guest argues that the next trade will be set by what happens in 2026 for 2027 and that dips in these names are attractive entry points.

Key Points

  • The guest argues that AI models getting closer to training themselves is the trend to watch, and that restricting a model for cybersecurity reasons can make it stronger.
  • Memory and compute stocks, which performed well year to date, pulled back over 40 percent at one point, which he sees as a good entry point after a momentum selloff.
  • Optical technology is growing in three use cases: scale-out, scale-across between data centers, and scale-up inside servers, with the scale-up ramp expected in 2027 and more so in 2028.
  • Nvidia benefits from open source models today, but over five to ten years many closed models may run on their own compute.
  • Anthropic's revenue growth is described as roughly 1,000 to 1,300 percent, with the key open question being when it reaches real profitability.
  • The guest says a roughly 2 trillion dollar valuation for Anthropic (around 196 times earnings) could be justified by longer-term potential, and that SpaceX's valuation is more about potential than Anthropic's.

Details Worth Noting

  • Concrete instructions: The guest gives no step-by-step buying instructions. His actionable guidance is to position now in optical and compute-related names during the pullback, watch what drives the next trade for 2027 as Q4 begins, and treat optical sell-offs on in-line numbers as a buying opportunity, since he says the sector is not one technology replacing another.
  • Named names to research: Coherent, Lumentum, Credo, Micron, Nvidia, Palo Alto (which the host says downgraded some names).
  • Anthropic valuation discussed at roughly 2 trillion dollars, with talk of 3 trillion; the guest says the key data to watch is profitability, ownership versus trading, and time to real profit.
  • Guest framing of AI models: the capability that makes a model dangerous for cybersecurity is the same one that makes it more useful, creating a double-edged sword.
  • The transcript is conversational and includes host and guest market opinions; these are not professional investment advice and the figures cited are as stated in the video.

Bottom Line

The video's core message is that AI infrastructure, particularly optical networking and compute, is still early, and the guest sees recent pullbacks as opportunities, but the valuation and profitability questions around private AI leaders like Anthropic remain open.