This bill is the entire American energy debate in one screenshot.
$5.88 to generate the electricity. $44.40 to move it 30 miles. The supply is 12% of the total charge. Delivery is 88%.
This ratio would shock most people, but Eversource customers in New Hampshire have been living it for years. And the gap is widening everywhere. According to the EIA, utility spending on electricity delivery rose 65% from 2010 to 2020 in real dollars, while spending on power production dropped 32% over the same period.
The reason is wild when you see the numbers. 70% of U.S. transmission lines are over 25 years old. 70% of power transformers are past 25 years. 60% of circuit breakers are over 30. The American Society of Civil Engineers gave U.S. energy infrastructure a D+ grade. Replacing the whole system would cost an estimated $5 trillion. Capital investment in distribution infrastructure alone hit $50.9 billion in 2023, up 160% from 2003.
And here’s what makes this politically toxic: every technology that promises cheaper energy generation, nuclear, solar, wind, runs into the same wall. The generation gets cheaper. The delivery gets more expensive. New Hampshire has a nuclear plant 30 miles from this guy’s house producing some of the cheapest electricity in the country, and it barely matters because the wires, poles, transformers, and substations between the plant and his outlet are aging, expensive, and regulated by a system that lets utilities earn guaranteed returns on infrastructure investment.
The more they spend on the grid, the more they’re allowed to charge. Utilities earn their profit from the delivery side. Eversource passes through supply costs with zero markup. But delivery? That’s where the regulated rate of return lives. Every pole replaced, every transformer upgraded, every mile of wire buried becomes an asset the utility earns a percentage on for decades.
That’s why this bill looks the way it does. The customer is paying 12 cents to keep a nuclear reactor running and 88 cents to maintain a grid built during the Eisenhower administration.
Three days between writing about fragile infrastructure and watching it break in real time. The cloud's concentration isn't a feature — it's a design flaw we've been deferring.
The question I keep coming back to: what would truly distributed compute actually look like?
Last Friday, I wrote that "the digital world is colliding with the physical one."
This morning, AWS reminded everyone what that looks like in real time.
A DNS issue in a single region took down airlines, banks, media sites, crypto exchanges, and government portals — even parts of AI infrastructure. One operational hiccup, and a third of the internet caught a cold.
The cloud isn't weightless or boundless. It's centralized, fragile, and running on physical systems that don't scale at the same rate as the abstractions we've built on top of them.
What we saw today isn't a bug — it's a symptom.
The question isn't whether we can make these systems bigger. It's whether we can make them smarter.
The next era of compute won't just be about capacity — it'll be about topology.
This has been on my mind for a while. The deeper I go, the more it feels like the next leap in cloud isn’t about software at all — it’s about where and how it’s built.
Curious how others in infra and AI are thinking about this collision between digital and physical.
More to share soon
The digital world was never separate from the physical one — we just built enough abstractions to forget.
A few years ago, I was deep in the trenches of DevOps — scaling systems that were supposed to be elastic, infinite, and on demand.
But the deeper I got, the more I realized how much of our world runs on layers we never think about. Every tool and pipeline sits on a deeper stack — one made of steel, power, and land instead of code. Behind every abstraction was infrastructure — and beneath that, the physical reality we've spent decades trying to forget.
We treat the cloud like it's weightless — as if it lives everywhere and nowhere at once. But it doesn't. It's racks and megawatts, bound by the same physics as everything else.
Peel back those layers and you see how precarious it all is — duct tape on duct tape, abstraction on abstraction, holding together a system barely keeping up.
Now the cracks are showing. Cooling systems fail in the same heat our models predicted. Power grids bend under the weight of our own algorithms. We built intelligence faster than we built the infrastructure to hold it.
The digital world is colliding with the physical one.
For decades, we tried to erase place. Now, we're being reminded how much it matters.
Maybe the next evolution of the cloud isn't about building bigger, but building smarter — more distributed, more adaptable, and closer to the energy that powers it.
The next generation of cloud won't just be written — it'll be built in places we stopped looking.
Just walked into @HobbyLobby expecting a totally different vibe. If I wanted overpriced crafts and homegoods I would’ve just gone to Michaels. What ever happened to cool storefronts that offered RC planes and other fun actual hobby items??
@zdrks@BioBayes@bryan_johnson@ESYudkowsky @NickBostromPhD True, but if ASI arrives within our lifetime, surviving long enough to influence its trajectory still matters. /dd isn’t about outliving ASI - it’s about being here to help align it
@zdrks@BioBayes@bryan_johnson@ESYudkowsky There is alignment because longer human lifespans accelerate intelligence gains and scientific progress, which @NickBostromPhD outlines as a potential path to superintelligence. If cognitive enhancement speeds up AGI, then /dd isn’t just survival - it’s an alignment strategy.
@jamesmchugh@bryan_johnson@anothercohen Checks out. Like others have pointed out, if natural production is sufficient - it doesn’t seem necessary. My personal take is the modification of exogenous testosterone just becomes a headache to manage especially when approaching levels that require an aromatase inhibitor
@jamesmchugh@bryan_johnson@anothercohen Further, correct me if I’m wrong, but I think TRT seems to raise resting heart rate along with excessive red blood cell count in some individuals
@jamesmchugh@bryan_johnson@anothercohen My understanding is that @bryan_johnson removed his caloric deficit and natural T levels returned making the need for exogenous test no longer necessary. Plus I think fertility biomarkers are essential for determining whether a longevity protocol is effective
I don't think many people understand what is happening in software development right now.
I have a few friends with computer science degrees. Yesterday I asked them how they use AI. One said he uses ChatGPT “a little bit.” The others criticized AI and basically were in denial of how good it's become.
Riddle me this:
How does a guy who looked at his first line of code last year build an app in a week, by himself, that would’ve required a whole team and several “sprints” a few years ago?
I sit at dinner with friends and family. All chatter about politics and pop culture. I bring up AI and get blank stares. Not one person has even heard of Claude.
The average person has barely used AI and has no idea what is happening.
I literally can't sleep at night.
Too many ideas. Too many opportunities.
I'm so excited.