WAHA premium and MAOP limitations
Interesting sections on energy in Anthropic's S1 filing!
There's been speculation on how data centers are managing fuel price risk from gas offtake, whether off-grid data centers will happen at scale
- 'an increasing share of data centers being developed with dedicated onsite generation independent of the grid'
- ANT exposed to fuel price volatility as they bear cost on a pass-through basis, no cap on fuel costs
- Lack of firm gas supply can delay facilities
- For grid connected assets, ANT discusses 'investing in curtailment systems' to reduce strain during peak periods
- Cooling needs may increase over time for data centers due to changing climate
- Notes risk that utilities may reallocate capacity required by ANT data centers
Austin Storms retweeted
The usefulness of Amazon’s 6-page memo has been nuked by AI-generated text (employees sit in meetings suspecting text is AI and tune out).
Suspicious memos tagged as AI;DR (AI, didn’t read).
Austin Storms retweeted
NVIDIA Vera Rubin NVL72 is available on CoreWeave.
@Cognition is running @devindevelopers in production on it, at up to 4.8x the total token throughput of GB200 NVL72.
V100 in 2017. Vera Rubin today. Same platform, every generation. crwv.co/utcq5
Austin Storms retweeted
had a blast with the galaxy team at galaxy stadium yesterday. loved seeing texas tech win their home opener in the newly renamed stadium.. a great way to end a great week
big things are being built in west texas
Austin Storms retweeted
We’re pleased to see @galaxyhq recognized by Governor @GregAbbott_TX among the companies committed to meeting Texas’ data center standards. We agree that reliable power, protected ratepayers, responsible use of water, and a strong focus on mitigating the impact on daily life for neighboring residential property owners and communities are the right priorities.
These are the ideals $GLXY built Helios around from day one: we fund our own grid infrastructure, use closed-loop cooling that minimizes water consumption, and we're proud to be Dickens County's largest employer and taxpayer.
We look forward to working further with the Governor’s administration to support responsible and transparent data center development across Texas.
gov.texas.gov/news/post/gove…
Austin Storms retweeted
We're proud to announce our partnership with Communities In Schools of the South Plains that will bring full-time student support services to Dickens County, Texas, home of our Helios data center campus.
Up to 85 at-risk students each year will receive one-on-one support, and every student in participating districts will gain access to program resources spanning college and career readiness, financial literacy, and family engagement.
This partnership builds on Galaxy's ongoing commitment to being a long-term partner in the communities where we operate.
Water security is quickly becoming infrastructure security.
Water-System Hacks Hit 7 States This Week, FBI Warns, As Trump Shoots Down Iran Theory zerohedge.com/political/wate…
Austin Storms retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
Austin Storms retweeted
Galaxy Stadium is the new home of Red Raider Football. 🏈
@TechAthletics is a storied program, from a Big 12 title to one of the premier home-field environments in college football. As the official data center and digital assets partner of the Red Raiders, we're proud to be part of that story.
We've been building and investing in nearby Dickens County, where @galaxyhq's Helios campus now carries 1.6 GW of capacity powering the infrastructure behind the code economy. Texas Tech grads are already on our team there, and for the next generation of talent, this is a direct pipeline into the AI and digital assets industry.
#WreckEmTech
Austin Storms retweeted
We’re excited to share Galaxy is now the official data center and digital assets partner of @TechAthletics under a 15-year agreement, renaming the home of Red Raider Football as Galaxy Stadium.
Texas Tech is central to our community footprint in West Texas. The university has one of the strongest talent pipelines in the country with Texas Tech graduates already working at Helios today.
This landmark announcement represents the first milestone in a broader strategic relationship rooted in digital innovation, including the exploration of potential projects to advance academic and commercial applications of artificial intelligence, expand research and workforce development, and catalyze economic investment across West Texas.
Phase 1 (133 MW critical IT load) is live at our Helios data center campus and generating revenue on schedule and on budget!
Execution is the differentiator in this business. Phase 2 (260 MW) is next.
The $GLXY team is building big things…
Phase I of our Helios data center campus is done. On schedule. On budget.
All 133MW of critical IT load now generating full revenue.
Phase II is already underway. 260MW more, live in H1 2027.
Big things coming.
$GLXY
Designed and built a very similar architecture at home ~3 months ago with ollama serving locally on a few Mac Studios and DGX Spark running under my desk.
Lots of tinkering happening by engineering-first organizations to rein in $/useful token spend.
How to keep AI spend flat while token usage grows exponentially: Not with friction and spend alerts. With better defaults, routing, and caching.
Better Defaults (not Usage Caps) – Engineers can choose any model they want, but defaults matter. We’re experimenting with defaulting to open weight models like GLM 5.2 and Kimi 2.7 through our LLM gateway, while still encouraging engineers to choose the right model for the task. 91% of our employees were never hitting their usage caps, so instead of lowering caps and driving up alerts, we're moving to cheaper defaults. Note that code reviews use a diversity of models, so they can check each other's work.
Better Routing – In our custom harnesses, we preprocess prompts and route to the best model for the job, considering cache hits and model pricing. For instance, you may want a frontier model for planning, but not for execution where they can be overkill. Ultimately, humans shouldn't be choosing models - AI can automate this task.
Better Caching – Cache misses are the easiest way to drive your cost up. All of our requests are cache aware, so we’re reusing a warm cache wherever possible. For example, our cache hit rate went from 5% → 60% in LibreChat once properly implemented.
Keep Context Lean – Start fresh sessions when switching tasks. Scope file context narrowly. Disconnect unused tools. Don't just compact. The goal isn't fewer tokens used, it's fewer tokens wasted.
Better Visibility – Our engineers can use as many tokens as they want, from whatever model they want, but we’ve made usage visible – and the more you spend on AI, the more impact we expect.
The goal isn't to suppress usage. It's to build the infrastructure that makes exponential growth sustainable.
Putting this into practice has cut our AI spend nearly in half, while our token usage continues to grow.
Austin Storms retweeted
Water usage has been a hot topic in the AI data center world, but the numbers may surprise you.
According to the Manhattan Institute, data centers use 0.2 percent of daily water usage in the U.S. and that number has dramatically decreased in the past few years due to a new method: liquid cooling.
By moving to 45°C liquid cooling, AI factories in favorable climates can use dry coolers instead of conventional cooling-tower-based systems, cutting facility cooling water use from roughly 2.6M gallons per MW per year to near zero.
Liquid cooling enables AI factories to be both water and energy efficient, while creating opportunities for heat reuse and dispersal to local communities, allowing these factories to become energy grid assets.
Learn more below ⬇️
nvda.ws/4gAVGoJ
Any technology that improves the ability to manufacture intelligence more cheaply, predictably, and at scale will likely find a place in the AI infrastructure stack.
