David Peterson

@edzitron If they decide to write off this equipment, and chuck it away, does that mean second-hand markets might get flooded with networking gear / drives / GPUs / etc ?

I think that if data center construction doesn't speed up dramatically in the next 6 months, hyperscalers will see massive write-downs on aging Blackwell GPUs as Vera Rubin ships. Either way, depreciation charges are likely to skyrocket in the next year. wheresyoured.at/premium-what-i…

Today’s premium newsletter is part 1 of my What If...We're In An AI Bubble? series, covering scenarios that could burst the bubble starting with token-based billing's destruction of most AI products, and the slow collapse of data center construction. wheresyoured.at/premium-what-i…

@PhiloGroves Peter is just showing us an early glimpse into our future. One day that won't be a $1.3M bill, it will be a $13K bill, then eventually it will be a $13 or even 13c bill.

We need to talk about token spend creep… $1.3 million per month cannot be efficient. The future of software should not add a $20,000/month fee to your business just for tokens to run a single feature, especially as you add hundreds of features…

@AxBerger @krishnanrohit He didn't, he painted images.

@krishnanrohit If only Claude could generate images...

In hindsight Shl0ms really should've disclosed upfront that the painting was done by Claude

@steipete > "How would we build software in the future if tokens don't matter?" You're basically giving us an early insight into what the world will be like in 3, 5, 8, 13, or 21 years from now? As costs are only going to go down!

People freaking out over my AI spend. What nobody sees: Part of what excites me so much about working on OpenClaw is that I'm trying to answer the question: How would we build software in the future if tokens don't matter? We constant run ~100 codex in the cloud, reviewing every PR, every issue. If a fix on main lands, @clawsweeper will eventually find that 6 month old issue and close it with an exact reference. We run codex on every commit to review for security issues (as it's far too easy to miss). We run codex to de-duplicate issues and find clusters and send reports for the most pressing issues. We have agents that can recreate complex setups, spin up ephemeral crabbox.sh machines, log into e.g. Telegram, make a video and post before/after fix on the PR. There's codex that watch new issues and - if it fits our documented vision well, automatically create a PR of it. (that then another codex reviews) We have codex running that scans comments for spam and blocks people. We have codex instances running that verify performance benchmarks and report regressions into Discord. We have agents that listen on our meetings and proactively start work, e.g. create PRs when we discuss new features while we discuss them. We build clawpatch.ai to split all our projects into functional units to review and find bugs and regresssions. We do the same split for security with Vercel's deepsec and Codex Security to find regressions and vulnerabilities. All that automation allows us to run this project extremely lean.

@Dieg12123 @Mattlo_notdead Back again now? First time though I've seen it glitch out that bad! And I'm a daily user.

@MathmoKiwi @Mattlo_notdead I think it's down too :(

se cayó Claude. no? :'(

@ya_man_Gd77209 Feels like it is down for me too.

あり？Claude落ちてる？まあしゃーないか。

@EricCrampton Aside from a small handful of countries who have stayed more or less unchanged, it seems that every single country has gone backwards since 2012??

With the current curriculum changes, the incoming cohort of 25 year olds should start improving in maybe a decade, full effect in 20 years. Until then, the cohort of 25-64 year olds will worsen as those who had been taught properly age out of the sample. Grim.

@ponekeben AT doesn't know what the meaning of the words "aim" or "low" or "costs" is.

$5k for a heritage impact report on some stairs for a wharf but definitely no need to change heritage rules according to Wellington's esteemed heritage advocates.

Will Smith eating pasta is now the official benchmark for AI progress. This is actually insane.

@lucasMnts @CWood_sdf With code you can write it with sufficient detail ;-)

@CWood_sdf good luck writing a sufficiently detailed specification on the first place

i love how people are saying "if we write a sufficiently detailed specification, the agent can write all our code" do you know what writing a sufficiently detailed specification that deterministically maps to what a computer's actions is? it's coding

Microsoft just gave Anthropic the same enterprise deployment access inside Office as Copilot. Microsoft is OpenAI's biggest investor. Claude for Excel, PowerPoint, and Word are now generally available. Outlook is in public beta. GA is the line where an add-in clears procurement and gets pushed through the same admin panel as Copilot itself. Anthropic just crossed it. The math on that deployment surface is brutal. Microsoft 365 sits at over 450 million paid commercial seats. Copilot, after 2.5 years of full-court selling by Satya, hit 20 million paid seats last quarter on the FY26 Q3 call. That's 4.4% penetration on the most aggressively marketed AI bolt-on in enterprise software history. The other 430 million seats are now reachable by Claude through the same IT deployment surface, the moment a CIO signs the manifest. The cross-app context is the second piece. A model that holds the same conversation as you move from Excel to PowerPoint to Word to Outlook eliminates the most expensive friction in AI productivity work: re-prompting every time you switch apps. Microsoft Copilot's "Work IQ" was the entire premise. Anthropic just shipped parity on that mechanic on Microsoft's own turf. The strategic paradox is that Microsoft pays OpenAI billions, runs Copilot on their models, then signed off on a competing AI getting GA-grade integration across Excel, PowerPoint, Word, and now Outlook beta. Why allow it? Microsoft's real product is the platform. Lock Claude out and you invite antitrust attention on every Office add-in. Let Claude in and Office gets stickier than ever. Microsoft wins either way. OpenAI's integration moat inside Office quietly became a tie. Citadel signing on as a launch customer for Claude for Excel signals where serious finance work is heading. Vals AI's Finance Agent benchmark currently has Opus 4.7 at 64.37%, the top score in the category. Buy-side analysts will install whatever scores highest on the modeling task they actually do, regardless of what their CIO is paying Microsoft for Copilot. The most valuable real estate in productivity AI was always going to be the cell next to your DCF. Anthropic just got cleared for full enterprise deployment.

@nbaschez Just a few simple pages, hand-written, linked together, and shared such that anyone in the world could open in a browser. I will never forget his answer… 'We can't, we don't know how to do it.'"

@nbaschez My father-in-law was a web dev. Sometimes it is difficult to get his attention when he is looking at old source code, because he is lost in wonder. We were browsing an archived GeoCities page one day, and I asked him what it would cost to build something like it today.

Would be cool if there was an easy way to share HTML pages You could imagine a world wide web of these documents, linked together Really cool idea

@amaanbuilds "what questions do you have before starting?" I agree completely; this was such a game changer, as I spend so much effort thinking about giving proper context to a LLM, but why had I never asked the LLM itself what extra context it thinks it needs?

@claudeai Got to give this beta a spin!

Claude for Excel, PowerPoint, and Word are now generally available, and Claude for Outlook is in public beta. As Claude moves between your Microsoft apps, it carries the full context of your conversation.

@luongnv89 @trq212 I agree, for anybody who isn't an HTML Maximalist, this is a nice middle ground to go for. While rapidly iterating stick with markdown, as you get closer to the final thing, get it back in HTML instead.

have u make a comparison between number of tokens for the same information, between markdown vs html ? i agree on the data structure and visualisation, but token is also matter a lot for the final out put: presentation, report - yes but for planning and mid process docs, markdown still the best option

@trq212 @somyatwts "HTML Maximalist" is detected ;-)

@somyatwts idk I think HTML is strictly better for all of that too

@trq212 I do the same, for the last year or two I've been asking LLMs to give me an HTML output, it just looks nicer.

@mcvrn @trq212 Got to ask @elonmusk to support HTML on X! (will turn X into the modern day Myspace??)

@trq212 “I’m don’t like reading long markdown files.” Writes a loong markdown x post.