OpenAI is holding preliminary talks with Amazon for an investment over $10 billion, including use of Amazon Web Services’ Trainium chips for AI training. This is separate from raising up to $100 billion at a $830 billion valuation, up 66% from October’s $500 billion mark. The $100B raise tells you what Sam Altman and OpenAI …
Ark Invest thinks Google has taken the lead with Gemini 3 + Nano Banana Pro. Brett Winton thinks we are heading from transformer architecture to a mix of transformer, diffusion and memory. In the medium term (1-2 years) Google’s distribution, bundling, and cost advantages are seen as the biggest structural moat. They predict that when …
In LMArena, Grok4.1 (Thinking) and Grok4.1 ranks first. In the earlier benchmark tests, Grok4.1 (Thinking) ranked first with a score of 1510. Currently, it is still first but with a score of 1483. Grok 4.1 is second. There is a massive reduction in hallucination. It drops from 12% to about 4%. This version scored more …
OpenAI has committed to spending $1.15 trillion (updated to $1.4 trillion) on hardware & cloud infrastructure between 2025 & 2035. OpenAI announced they would hit $100B in 2027 which is earlier than expected. OpenAI expects to have a $20 billion per year revenue run rate at the end of this year (2025). If OpenAI hits …
David Sacks, AI and Crypto Czar, said there will be no federal bailout for AI. The U.S. has at least 5 major frontier model companies. If one fails, others will take its place. OpenAI CFO Sarah Friar said OpenAI wants the U.S. government to provide a backstop or guarantee for AI financing. Sam Altman believes …
In the USA, AI data centers will need around 20 to 30 gigawatts (GW) of combined power by late 2027. 30 GW is about 5% of US average current power generation capacity and 2.5% of China’s. 30 GW would be around 25% of power generation capacity in Japan, 50% in France, and 90% in the …
2025 saw a tripling of continual learning LLM papers according to arXiv trends. This is driven by foundation model scale and multimodal extensions. However, no flagship AI released models (GPT-5, Grok 4 etc…) fully integrate production-grade continual learning yet. There are expected architectures like routers and sparse finetuning signal imminent hybrids. There has been no …
Mathematician Thomas Bloom, runs erdosproblems.com, explains that OpenAI surfaced erdos problem solutions that were not unsolved. OpenAI GPT-5 simply surfaced existing papers solving them that he had personally overlooked, mislabeling them as “open” on his site. No new mathematics was generated—GPT-5 acted as an advanced search tool. OpenAI’s Mark Sellke and Sebastien Bubeck previously announced …
Brad and Bill on the BG2 Podcast describe the AI CapEx bubble. The explosive growth in AI investment—especially in data centers, chips, and model training—shows potential signs of historical overreach. They increase the odds of overbuilding and reduce the quality of revenues. Gurley identifies round-tripping deals like the original deal between Microsoft and OpenAI, where …
Broadcom and OpenAI announced a multi-year strategic collaboration to co-develop and deploy 10 gigawatts (GW) of custom AI accelerators and integrated networking systems. This is OpenAI’s start at designing its own AI processors. This builds on the Nvidia and AMD GPU supply deals. Broadcom’s stock is up about 10% (adding ~$150 billion to market cap) …
AI first companies are generating real revenue. The leading company is OpenAI and Anthropic also has significant revenue. There is also very fast growth in hundreds of small companies and that revenue growth is faster than SAAS (software as a service). XAI did not have much revenue in 2024 but it has made substantial gains …