Keep pulling the thread on Steve Jang.
Open-source AI models developed in China are gaining significant adoption among U.S. enterprises, on the condition that they are hosted within the United States.
Large, deep reasoning models from companies like OpenAI and Anthropic require 150 times more compute resources than smaller AI models.
AI compute has become the 21st century's key strategic resource, analogous to the role oil played in the 20th century, and is essential for global competition.
A successful wave of initial public offerings from late-stage private companies like SpaceX, OpenAI, Anthropic, and Databricks is needed to unlock capital and ease the liquidity pressure on early-stage venture capital funds.
NVIDIA currently holds a market share of over 90% in the AI and machine learning accelerator chip market.
Steve Jang predicts that NVIDIA's dominance in the AI chip market will face increased competition as the market shifts from training-focused to inference-focused workloads.
In the current year, the balance of AI compute usage is shifting, with inference workloads expected to grow from one-third to two-thirds of all compute.
Generative media startup Fowl has raised funding at an $8 billion post-money valuation.
Perplexity is the first company to have launched complex, end-to-end AI agents to a mainstream audience.
The primary advancement in robotics over the past few years has been the integration of intelligent reasoning models, enabling the creation of truly adaptive robots.
In most technology sectors, the market ultimately consolidates to three or four dominant companies by the time the sector matures.
The three leading frontier AI model labs in the Western world are OpenAI, Anthropic, and DeepMind.