@jukan05 There’s going to be conditions on the purchase
The startup needs committed contracts to build and scale. They can take this commitment and get financing now
English
Daniel Youssef
384 posts
Daniel Youssef
@DanYouHQ
Tracking AI from the land, power, chips, to models. Dealmaker and executor of AI infrastructure. Principal, Infrastructure Partnerships & Deals @AWSCloud
New York, NY Katılım Nisan 2009
1.2K Takip Edilen264 Takipçiler
Datacenters in space are making more and more sense.
Garry Tan@garrytan
Per one-gigawatt data center complex: 5,322 permanent jobs, $157M per year in state taxes, $248M per year in local taxes. During construction: $2.67B in combined investment But nobody managed to tell Seattle this, so it's banned.
English
@TheStalwart That’s why A2A protocol now exists. Everything we use now is optimized for human use with a GUI…that’s def going to change
English
It seems obvious that this is, at least to some extent, the future. Internet designed for bots with much more metering of use and consumption.
squeakerlord@squeakerlord
@TheStalwart Lllms need to cough up the cloudflare pay per crawl tax or you’ll start seeing more and more of this.
English
Daniel Youssef retweetledi
@jiahanjimliu Every cloud provider already changed their cpu lifecycle to 6+ years. It was a big deal a few years ago when they did it
English
For Neoclouds, the biggest news today is that $META is extending even non-AI servers to 6-7 year lifetime.
GPUs use an order (10x+) more HBM than CPUs. If CPUs are seeing 6-7 year lifetime, GPUs will see 7-8+ year lifetimes.
Adding 2-3 years is going to be pure profit on $IREN, $CRWV, $NBIS bottom line with basically zero incremental capex.
Wall St Engine@wallstengine
$META is extending the life of some non-AI data center servers from 6 years to 7 as a memory chip shortage creates a “significant server supply deficit.” The shortage is expected to last through 2027, according to internal memos viewed by the Journal.
English
Powell staying on as a central bank governor is objectively funny
English
The entire industry is moving away from $/GPU-hr
CoreWeave@CoreWeave
$/GPU/hour doesn't tell the whole story. If your AI cloud isn't purpose-built, you're paying for idle time. A thread on the 5 factors for true AI TCO 🧵
English
i’m sorry, but anthropic and openai are not $1T companies
shirish@shiri_shh
The biggest IPO run in the history of the market three $1T+ companies. possibly all going public in the next 12 months. SpaceX IPO: $1.75T. OpenAI IPO: $1T Anthropic IPO: $1T we're living through the greatest technological wealth creation in history.
English
@TheStalwart Macs selling because of local LLMs don’t move the needle for Apple on revenue and they aren’t deliberately marketing the macs that way
English
This might a dumb question but why do people talk about Apple not being a player in AI at the same time as people talk about Macs instantly selling out because people use them for self-hosting LLMs?
English
@ASpittel Some of the other internal tools are bigger game changers 😇
English
getting access to quick desktop internally was i think my biggest "everything's different now" moment i've had in a while.
#quickaf
English
FERC reviewed 3,500+ pages of public comments on data center interconnection and still can't hit April 30. Punted to June. The grid is not keeping up with the demand signal and the regulatory process is the bottleneck, not the utilities
English
@AndrewBenson Except databricks isn’t a SaaS platform like what you’re comparing it to lol there’s a reason it’s best positioned for the AI revolution
English
Databricks is possibly in the worst possible position for IPO. They would enter the SaaS market segment right now. Best pivot is probably to build out their own datacenters and become a neocloud. Hard to hang on to the valuation this year.
English
Crazy as it sounds Nucor might be the best of the undiscovered data center plays!
English
👀OpenAI leaders believe they can generate a lot more money from you if you pay ChatGPT $12 ~less~ per month but agree to start seeing ads…
English
I use a lot of different AI tool but creating #skills, context, etc between all of the tools is a major PIA. What setups do people use to retain context, skills, etc across various platforms? @AnthropicAI @OpenAI @perplexity_ai
English
Daniel Youssef retweetledi
Anybody who has tried @OpenAI's GPT 5.5 sees its truly a frontier model. They've locked up a lot of compute. The shakles have been taken off and now they can be hosted on @awscloud and @googlecloud . They've been making very strategic senior hires in the enterprise sales space. Like it's wayyy too early to count them out
English
the only interpretation on UAE leaving OPEC is more production which drives down oil prices over the long term. UAE clearly wants to be the middle east big dog
English
800 VDC is not a future thing, it is happening now. GPU density is so high the power conversion literally can't fit in the rack anymore. Schneider's CTO Jim Simonelli called it at Data Center World 2026. Sidecar and centralized distribution are the new normal. The whole rack form factor assumption is getting ripped up
English