Not terrible

Introducing Project Genie: An experimental research prototype powered by Genie 3, our world model, that lets you prompt an interactive world into existence — and then step inside 🌎

Quick notes on Microsoft $MSFT

Apptopia: ChatGPT's US market share fell from 69.1% to 45.3% between Jan. 2025 and Jan. 2026; Gemini rose from 14.7% to 25.1%, Grok rose from 1.6% to 15.2% (@kantrowitz / Big Technology) bigtechnology.com/p/new-data-ope… techmeme.com/260203/p27#a26… 📥 Send tips! techmeme.com/contact

We’ve started to measure time horizons for recent models using our updated methodology. On this expanded suite of software tasks, we estimate that Gemini 3 Pro has a 50%-time-horizon of around 4 hrs (95% CI of 2 hr 10 mins to 7 hrs 20 mins).

One of the factors that led Germany to invade Belgium was the idea that Russia was industrializing quickly and would overwhelm them in a few years. Is Taiwan trying to stay in the "goldilocks zone"? Build military infra too fast, CCP feels like its now or never?

I don't have access to sell side reports.. is this Azure model (assuming its driven by supply) from SemiAnalysis still directionally correct?

The way to interpret these ads is that Anthropic is giving up in consumer.. but in exchange for a strategy credit, leaning into enterprise.

We estimate that GPT-5.2 with `high` (not `xhigh`) reasoning effort has a 50%-time-horizon of around 6.6 hrs (95% CI of 3 hr 20 min to 17 hr 30 min) on our expanded suite of software tasks. This is the highest estimate for a time horizon measurement we have reported to date.

You can think of workflows in enterprises as "inverted trees". You move down from the trunk, to branches, to leaves, and leaf veins. Ever increasing in the level of detail in the abstractions. The purpose of vertical SaaS (and defensibility vs. labs and horizontal SaaS) is to

Gemini was the fastest-growing Gen AI tool in January 2026, followed by Claude and Grok.

Internal memo: Sam Altman tells staff that ChatGPT is "back to exceeding 10% monthly growth", and OpenAI plans to release "an updated Chat model" this week (CNBC) cnbc.com/2026/02/09/sam… techmeme.com/260209/p17#a26… 📥 Send tips! techmeme.com/contact

Yossi Dahan Noam Brown ARC-4 is in the works, to be released early 2027. ARC-5 is also planned. The final ARC will probably be 6-7. The point is to keep making benchmarks until it is no longer possible to propose something that humans can do and AI can't. AGI ~2030.

Lots of "zombie SaaS" incumbents that look like value but are hollowing out. Typical arch looks like this: - They "bolted on" AI features to existing platform, don't have talent to rebuild from first principles - AI is rising tide, NRR up as customers pilot AI features - GR and

Excluding an absolute melt down, it seems likely that if OAI or Anthrophic got to $800B in revenue in ~5-7 years, they'd be able to raise $200B to cover a demand and supply gap, no? Also, my understanding is that most of the OAI commitments are in tranches based on milestones.

Who is more likely to 10x from here, one of the Mag 7's or one of the below investments?

If your a native AI start-up, do research on what renewal cycles look like for your incumbent competitors (e.g. do they co-terminate contracts Dec. 31st). If they are dumb enough to do price increases, you can *aggressively* take market share away if you time things correctly.

Given Seedance 2.0, its possible this happens this year. Similar to Napster, Limewire, etc. - if owners of IP don't figure out a legit way to empower users and creators on their own terms (ex. Spotify), piracy will explode.

Why did this never transfer over to the Japanese making great software?