PJM at FERC Monday: one-year large-load study, ~2x faster, plus a rule that big compute loads ride through grid disturbances.
Speed to power now ships with an uptime duty.
Who owns ride-through: utility, developer or tenant?
utilitydive.com/news/pjm-rto…
xitter.cf/moeeev/status/21216267…
4
Blackstone's BXDC agreed to buy 75% of 13 AWS powered shells worth $5.0B: 647 MW, 7.0% cap rate. That's ~$7.7M per MW before a single GPU.
My read: the concrete isn't the asset. The energized grid connection is.
What's a signed energization date worth?
sec.gov/Archives/edgar/data/…
2
ISO-NE's new large-load filing: non-firm data centers get re-dispatched whenever local power hits $250/MWh, about 1% of hours today.
That's uptime written as a contract term, not a generator spec.
What's your plan for that 1%?
isonewswire.com/2026/11/13/i…
xitter.cf/moeeev/status/21216267…
9
Latitude: 120 GW of data center load is signed, 22 GW is running. Signed isn't energized.
Grid operators answer FERC on large-load rules by Nov 16. That's where uptime gets defined: firm or curtailable.
Who underwrites the gap?
latitudemedia.com/news/quant…
xitter.cf/moeeev/status/21216267…
5
Malaysia plans to make data centers disclose GPU imports and power use to catch smuggled chips, and to charge for reserved capacity left unused.
My read: the second rule bites harder. Paper megawatts now carry a cost. Who drops reservations first?
tomshardware.com/tech-indust…
11
ERCOT says nearly 10 GW of medium loads sit at 60–74.9 MW, just under its 75 MW large-load review.
My read: that pile-up is the data. Bunching below a line is how economists price what crossing it costs.
If the line drops to 50 MW, where does the pile go?
voltgridnews.com/story/ercot…
9
Fukuyama on Carnegie today: AI hits screen work before electricians, and a UBI check can't replace the recognition a paid job confers. My read, via Kojève's Hegel: status may migrate to whoever moves atoms. Which jobs gain respect first?
carnegieendowment.org/podcas…
1
1
19
SoftBank booked a $3.7B loss on AI data center power gear it prepaid for, and its CFO says it bought before long-term specs were final. My take: capacity reservations are now a bet on your own spec sheet.
Who's writing smarter cancel terms?
cnbctv18.com/market/softbank…
18
Norway's grid operator Statnett is pausing data center power hookups to manage grid strain.
My take: cheap hydro was never the moat; deliverable megawatts are. If Norway has a waiting room, the "just build where power is cheap" plan has one too.
xitter.cf/ConstrucReview/status/…
Statnett Pauses Data Center Power Hookups to Control Norway’s Grid Strain constructionreviewonline.com…
12
A builder is archiving Portuguese charger status every five minutes because the national feed only shows the present.
My take: uptime needs a history, not a screenshot.
The useful question isn't 'is it online?'
It's 'how often wasn't it?'
xitter.cf/BlnaryMlke/status/2119…
🇵🇹 Portugal’s EV charging network (NAP) publishes live snapshots only, no history. We’re building AmpereArchive: a 5-min archive of every charger’s status as Parquet on @CloudflareDev R2 (no egress = perfect fit). Would @Cloudflare want to support it?
24
Anthropic took over someone else's Caterpillar order: 1.7 GW of gas generators for $3.4B, per the FT. About $2,000 per kW. My take: the real purchase was the delivery slot. In AI power, a two-year-old order beats a new idea.
Who sells theirs next?
investing.com/news/stock-mar…
21
The AI bottleneck isn't GPUs.
It's power that's already on — EV charging sites, flexible grid load, and home data centers.
1
28
Chip fundraising is loud right now, and mostly un-investable from here. Big checks. No lever I can actually pull.
The moat isn't the GPU. It's the energization date.
1
2
YC Liquid Energy launched NanoDC: factory-built solar+battery AI boxes that skip the grid queue. Deploy in ~4 weeks, on in 30 min. Interconnect is the new moat — startups are tunneling under it. Does this scale past demo?
xitter.cf/_LiquidEnergy/status/2…
We built a new type of datacenter for AI inference. We call it NanoDC.
NanoDC is a solar-powered autonomous datacenter that can be deployed anywhere on the planet in 4 weeks.
11 Trillion will be invested in AI compute over the next 4 years. Most of this will be for inference and most of it will be severely delayed: stuck in grid interconnect queues, site permits, or supply chain issues. NanoDC sidesteps these issues completely.
NanoDC is built rapidly at a factory and shipped to the site ready to be turned on in 30 minutes. It generates its own power with solar and batteries. You open the door and the solar panels fold-out like an accordion. The compute is cooled with patent-pending high-density air cooling. It connects to the internet via Starlink.
No construction. No grid interconnect. No Water supply.
It never waits for transformers and turbine generators either.
Getting more compute is as easy as dropping more units and turning them on.
The first units will be live this month. Thousands will be shipping starting 2027.
Sunlight in. Tokens out.





