
dev at @asana • big OSS enthusiast • team space 🚀 • pls try @analogdotnow
Poland
Joined December 2023
- Tweets3.8K
- Following417
- Followers190
- Likes20.4K
> Their five largest customers went from 36% of revenue in Q1 2026 to just 21% in Q3
Let me spoil this for you (don't fall off your chair):
- Cursor (SpaceX)
- Microsoft
- Google
- NVIDIA
- Amazon
Some interesting stuff from digging into @AnthropicAI's S-1 filing today:
- Claude Enterprise users who don't use Claude Code spend less than $75/month on average, but the top 5% who do use Claude Code spend roughly 20x as much
- They recently delayed pre-training a new frontier model for more than a month because they needed to improve security measures against model weight theft
- Over 80% of Anthropic's revenue in Q3 2026 came from model versions released within the preceding six months
- Anthropic expects to pay @SpaceX roughly $3.1B per month for compute once all contracted capacity comes online. They could effectively be paying $37B/year, up to $120.5B in total, but the contracts can be cancelled with 90 days' notice after early 2027
- Over 90% of contributions to Anthropic's own codebase in September 2026 were written by Claude. The average daily volume of code contributed per engineer in Q3 was 45x what it was in 2024
- Their five largest customers went from 36% of revenue in Q1 2026 to just 21% in Q3, but every single one of those five customers actually spent more in Q3 than ever before
- They recognized a $2B operating expense in Q3 just for matching employee charitable donations. It issued roughly 28 million shares in October to settle the obligation
- Their $42B Broadcom financing facility has an interesting clause. Normally, the notes carry 9% interest and convert into shares at a 10% discount, but if Anthropic misses certain revenue targets, the interest rises to 13% & the conversion discount increases to 15%
- Google has the option to purchase up to $10B of additional nonvoting Anthropic shares at the IPO price
- More than 11,900 customers now spend at least $100K annually with Anthropic. Over 1,900 spend $1M+ and roughly 250 spend $10M+
- Claude Cowork is seeing faster adoption than Claude Code did at the equivalent point after its launch
With all due respect, that’s a bot for eating tokens out of the usage limit.
Not everyone has a special unlimited sub
Want to know the latest in @bot?
I made a What's New Bot:
- Checks x.ai/bot/changelog each day
- Checks my latest what's new video
Delivers it straight to your chat
Get it here: x.ai/bot/dEFFB5_KX5U06lNJWVZ…
Dario today
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
































