@adiebald
Account based inGermany!
About this account
- Account based in
- Germany
- Connected via
- Germany App Store
X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
Tech Executive & Founder. Thinking out loud on AI strategy, energy, mobility & deep tech.
Munich
Joined July 2008
- Tweets806
- Following273
- Followers344
- Likes1.1K
Pinned Tweet
This is the right frame, the scarce asset is institutional judgment, not just data.
The question is which parts of the context graph (the captured decision traces) must stay fully sovereign, and which, if any, are undifferentiated heavy lifting, better pooled in a shared base layer.
A precedent exists in open source, where companies commoditize what doesn't differentiate and compete above it. AI is already running this playbook at the model layer (open weights) and the protocol layer (MCP). At the context layer it hasn't happened yet.
Not a bad approach, if member states actually agree.
Europe holds essential chip tech every fab needs (ASML, Zeiss, Trumpf).
And it can trade fast-permit datacenter sites for guaranteed frontier access.
Die Strategie: transformative-ai.eu/
One of the best takes so far.
Presto furioso on the public stage when all that’s needed is andante risoluto in the lab.
Joined @EdLudlow to talk about AI progress, independent evaluation, and why I’m optimistic.
Some quick takes:
- We’re better at building AI than understanding it. Attention towards testing/evaluation matters more than slowing down.
- Our RSI Index projects models could match human researchers on the tasks we test by August 2027. Embedded evaluators can produce more accurate estimates based on internal systems.
- Public conflict masks cooperation. The labs, policymakers, and enterprises we work with want better evidence. I’ve seen enough to believe coordination is possible.
- Independence comes at a cost. We’ve rejected contracts that would compromise ours. The same group doing the testing shouldn’t also sell the solution.
- Evaluation should scale through better technology. If it becomes a bureaucratic moat for incumbent labs, we’ve failed.
- Market-based evaluation has a role with or without regulation. Competition pushes us to build better technology and keep up with the frontier.
Celebrating the regulation of successful companies does not build European AI innovation.
We have designated ChatGPT as a Very Large Online Search Engine and Reddit and Roblox as Very Large Online Platforms under the Digital Services Act.
They now have four months to comply with additional DSA obligations.
More: link.europa.eu/8bmk3M
#DSA
ALT “Graphic with a purple background and yellow EU stars. Text reads: ‘ChatGPT, Reddit and Roblox designated under the Digital Services Act.’ The European Commission logo appears in the bottom right.”
Raspberry Pi is an early MHS adopter! Great signal for embedded + physical AI.
Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing.
Read more: anthropic.com/news/model-har…
Awesome - cheap, open, and actually a product: the Greenphone moment for physical AI.
BIG ANNOUNCEMENT FROM HUGGING FACE TODAY:
We're unveiling Microduck 🐥🤖
It's a tiny $399 open-source robot you can teach new tricks with reinforcement learning. It can walk, pick things up, get back up when it falls, and even roller-skate.
Welcome to the era of open-source affordable robots to democratize physical AI and world models!
🤗🤗🤗
The free stealth model ox-alpha (now GLM-5.3-Flash) ran on "Chinese AI chips" serving ~10T tokens/day on OpenRouter + OpenCode.
z.ai/blog/glm-5.3-flash
Awesome! Can’t wait to Waymo around.
Servus, München! 🇩🇪 We’re bringing the Waymo Driver to Germany. Over the coming weeks, our vehicles will arrive in Munich to begin laying the groundwork to launch our fully autonomous ride-hailing service in late 2027!
Read more: waymo.com/blog/2026/08/waymo…
Great report from ICONIQ! My highlights:
1. Vertical AI crowds into financial services & healthcare. Product & Design drops to #10 (p. 7)
2. 180 agent PRs, 0% passed CI. Speed and scale without verification obviously fails (p. 41)
3. Model selection criteria increasingly favor open weights (p. 11)
4. GTM and pricing are still exploratory (p. 17)
Our 2026 State of AI Report shows that 30% of AI companies bundle FDEs into the subscription. 29% charge a service fee. 25% run a hybrid.
That fragmentation isn't a pricing choice. It's a sign that the market hasn’t yet found a consistent way to price the value an FDE actually creates.
Read more: bit.ly/4g5wDse
Disclaimer: bit.ly/3H4dQj0
It's good, but who else needs a det-ox?
🚨 Ox Alpha might actually be AGI
I gave it and GPT-5.6 Sol (Ultra) the exact same prompt to build a website with cursor-reactive magnetic motion.
Ox Alpha delivered faster and, in this test, far more accurately.
the motion feels better, the graphics look cleaner, and even the font choices are way ahead of what GPT-5.6 Sol gave me.
I’ll drop the exact prompt I used in the comments for anyone who wants to try it.
Really impressive. In legal, post-trained open-weight models now match or beat the frontier at a fraction of the cost. Which vertical is next?
SoftBank invests in OpenAI.
OpenAI leases LPS from SoftBank’s energy company.
NVIDIA invests in OpenAI.
OpenAI buys NVIDIA chips and deploys them on that LPS.
NVIDIA backstops the LPS lease in return for exclusivity.
Some see circularity.
Others see a call option on OpenAI’s wallet share (or market share if OpenAI doesn’t use it).
Neither is wrong.
The kaiyuan point is the important one. Most open weights are permissive, not open source (opensource.org/ai). Nothing to contribute back to, more like freeware that spreads quickly. Which is the strategy.
The next China shock will come from open-source AI ft.trib.al/nik5Uuj | opinion
Unprecedented numbers from companies with strong balance sheets and large operating cash flows (disclosed in filings, unlike Enron).
The bet remains the same: enterprise app revenue has to arrive as the leases start, so that free cash flow does not turn or stay negative.
And the lower layers have to hold: neoclouds still able to refinance, and the chips achieving their estimated useful life.
Massive spending commitments for data-center leases and chips aren’t shown on companies’ balance sheets. on.wsj.com/4qEIFxL
Mistral is packaging its own models and third-party open weights into a European neocloud offering, financed by five-year customer commitments with no exit.
That lets customers keep their data and decisions, their alpha, in sovereign hands.
What remains open for customers: rewiring the business processes and defining a new split of work between people, agents, and robots.
venturebeat.com/infrastructu… via @VentureBeat
pp. 148–149 is the one. From May 2025 to April 2026, all human-feedback vendor traffic ran without blocking bio classifiers: ~50k people, ~133M exchanges, vendor-vetted only, mostly open-ended access. An internal-use flag disabled both blocking and the logging of classifier flags, so no flags reached review for a year.
The retrospective pass came back clean, 62 non-red-team transcripts flagged, no clearly concerning misuse on manual review and no customer impact.
Anthropic raised its CB-1 risk assessment on the back of it. What stays is the conclusion: this raises the likelihood of similar issues they haven't found.
As part of our Responsible Scaling Policy, we publish regular Risk Reports. These share detailed information on the risks of our systems and how prepared we are to address them.
Our second Risk Report is now available: anthropic.com/aug-2026-risk-…
$40 billion ARR, 2x this year, Codex hockey-sticking, and a CRO swap (CRM → Wiz) weeks after confidential S-1.
Looks like rebuilding the revenue mix before IPO.
OpenAI is on track to generate annualized revenue of more than $40 billion based on its current performance, according to sources, roughly doubling its run rate from the end of 2025 and bolstering the company’s plans for a Wall Street debut bloomberg.com/news/articles/…
