Scott Alexander: “An AI superforecaster is an AI – usually a frontier model like ChatGPT or Claude – which has been modified to be good at forecasting. This usually means a “scaffold” – a program that handholds it through a long research process with various prompts, tools, advice about when to create subagents, etc. The overall experience is a lot like using any other AI, but slower and more expensive, because it’s doing more work.”
WSJ: “Unlike other great powers, America’s strategic power has been rooted, from the beginning, in the dynamism of a society more commercial than governmental, more private than public, and more civilian than military, even in war. The U.S. advantage wasn’t in the White House, the State Department, or the Pentagon. It lives in a nation of joiners and problem-solvers who band together rather than wait for direction from the state.”
SaaStr: “Two things are true in B2B right now, and they look like they can’t both be true. Total software spend is growing 15% this year, the fastest in a decade, up from 12.8% last year. Gartner has it going from $1.2T to $1.4T. At the same time, public software is trading at a discount to the S&P 500 for the first time ever, and leaders like Monday, HubSpot, and Atlassian got cut 60% to 70% in a couple of months. Spend is accelerating. Yet for many software leaders, valuations are collapsing. Both at once. That’s the whole story of B2B in 2026, and it resolves the moment you stop looking at “software” as one thing. The market has split in two. One group is tapping AI budget and re-accelerating, in some cases to numbers we’ve never seen at scale. The other is running the same playbook from 18 months ago, waiting for a recovery that is not coming. There’s very little in the middle.”
Mint: “Open-weight models make their trained weights publicly available, allowing developers to download, fine-tune, and deploy them on their own infrastructure. However, the training data and training process are usually not released. Examples include GLM (developed by Zhipu AI), Qwen (Alibaba Cloud), Kimi (Moonshot AI), and DeepSeek (DeepSeek AI). In terms of cost, proprietary models are generally more expensive than open-weight models. For example, GPT-5.5 costs $5.50 per million input tokens, while DeepSeek-R1 costs $1.35 per million input tokens. Input tokens are text, numbers, or code that are fed into an AI model. Illustrating the difference further, Additi Upadhyay, co-founder of AI startup, Noveum AI, said, “Take a frontier proprietary model like GPT-4o versus an open-weight model like DeepSeek V3. GPT-4o runs about $2.50 per million input tokens and $10 per million output tokens (generated by AI models). For the same amount of tokens, DeepSeek V3 is roughly $0.27 for input and $1.10 for output. That’s close to a 9x difference for almost the same quality on a lot of everyday tasks.””
