Перейти к содержимому

Higgsfield's $1B Run Rate, Island Security, the Inference Boom & AMD Buys World Labs

The Cap Table — Pre IPO Podcast и ещё 2

0:00 / 0:00

Higgsfield's $1B Run Rate, Island Security, the Inference Boom & AMD Buys World Labs

244 просмотра · 2 нед. назад
The Cap Table — Pre IPO Podcast и ещё 2
244 просмотра · 2 нед. назад
The AI infrastructure layer most investors can't name is where we're focused right now. Higgsfield has crossed a $1 billion annualized revenue run rate about 18 months after launch, following an August Series B at a $5.4 billion valuation, and most of its revenue now comes from business customers rather than individual creators. Inference platforms are repricing quickly. Baseten is reportedly in talks at around $26 billion, Fireworks AI has reportedly considered raising at up to $30 billion, Fal has reportedly discussed $15 billion or more, and Modal is reportedly closing in near $15.75 billion. Against that backdrop, Together AI's July Series C at $8.3 billion, on more than $1.15 billion in annual bookings, stands out as relative value in the group. In cybersecurity, Island raised $400 million at a $6.4 billion valuation as enterprises race to govern AI agents inside the browser. AMD's roughly $8.2 billion all-stock deal for Fei-Fei Li's World Labs shows chipmakers moving closer to the model layer as physical AI and robotics come into view. Our house view: if most compute becomes inference and most inference runs on open models, the platforms serving those models can benefit whether the frontier labs keep accelerating or not. Aaron Ross and Aaron Dillon break down the week in private markets, from AI video to the infrastructure quietly powering open-source AI. In this episode: Higgsfield's path to a $1B run rate and the "AI wrapper" debate Why B2B moats look stronger than B2C, and where Higgsfield fits as it moves upmarket Inference platforms 101: Together AI, Baseten, Fireworks, Fal, and Modal Ross's thesis: inference platforms as a way to play both sides of the frontier-lab story Dillon's call: most compute will be inference, and most inference will be open source Island, AI agents, and how to think about cybersecurity exits and TAM AMD–World Labs, Decart, and whether chipmakers are positioning for robots 00:00 Intro: Higgsfield, Island, inference platforms, World Labs 00:35 Higgsfield's $1B run rate in 18 months 01:58 Hands-on with Higgsfield: democratizing Hollywood-level production 04:52 Is Higgsfield investable? The "AI wrapper" debate 06:16 B2B vs. B2C: where the moats are 08:07 On-demand movies made for your family 09:40 Inference platforms explained: Together AI, Baseten, Fireworks, Fal, Modal 10:50 Ross's thesis: inference as a hedge on the frontier labs 12:38 Dillon: 90% inference, 90% open source 14:25 A smaller slice of a much bigger pie 16:23 Can inference platforms be $100B+ companies? 18:59 Why Together AI stands out 21:14 Island and the overlooked cybersecurity trade 23:24 Cyber as AI infrastructure: exits, niches, and TAM 26:53 Stock selection, then AMD buys World Labs 28:18 Decart, world models, and a robotics-chip play 30:28 Full-stack AI: Unconventional AI and custom silicon 31:43 The robots are coming: Tesla Optimus and positioning early 33:07 Close This content is for informational and educational purposes only and does not constitute investment advice, an offer to sell, or a solicitation of an offer to buy any security. Opinions expressed are those of the speakers as of the recording date and may change without notice. Private company securities are illiquid, speculative, and involve a high degree of risk, including the potential loss of the entire investment, and are generally available only to accredited investors. The hosts or their affiliates may hold positions in, or have business relationships with, companies discussed. Valuations and figures cited are based on public reporting and have not been independently verified.