I totally agree with this. In my on the ground reporting in four different communities across the US over the past six months (Arizona, Texas, Michigan and Louisiana) I have been so surprised that AI data center developers don't consider throwing more money -- maybe a LOT more money -- at their problem of getting towns on board with their projects. treat these communities as the valuable resources they are, not a group to take advantage of with legal wrangling. It may not always work, but I think there would be a lot more support if the share were more valuable and fair.
I broadly respect the power of political pork, but (A) you'd need to be careful to not fall afoul of vote-buying laws with direct cash (B) village bribes may not circumvent larger state-level movements (C) presumably labs are cautious about starting a race to hell on welfare when they have S-1s to keep clean.
Maybe pinning the blame of anti-datacenter discourse on China will backfire. Or maybe it will be the only practical political lever remaining in 2027 to keep datacenters around. Exponential growth applies to populist movements too.
The diffuse-gains, concentrated-pain split is the right lever, but the concentrated cost is more concrete than general dislike. PJM's capacity costs jumped about 174 percent for the 2025-26 delivery year, and its own market monitor named data center load as the primary driver, which shows up directly in nearby residents' bills. A revenue-share check only flips a town if it clears that rate increase and stays visible.
Oracle, Palantir and other legacy big data center vendors are scamming the public, the press, that’s you - elected officials - that the future is massive energy-consuming data centers.
Any application requiring a data center, can be built with low I/O technology, and run on one or more Apple minis or Intel tiny computers, using less power than a Tesla, no water cooling, at 1/10th the cost and will run 100 to 1,000 to 1 million times faster - no data center needed.
Quantum computing companies are soon to be a dominant force in high compute problem solving. Trillions of dollars are being invested in quantum computing. Quantum computing does has no need for current data centers - none.
New technologies like virtual MESH are being demonstrated today, where the contents of an entire data center can run, instantly, on existing processors from a phone, tablet, server, Raspberry Pi - duplicating a data center with virtual no additional energy - or water - or environmental devastation.
A.I. applications are moving from LLM-centric “capture all knowledge” to where the money is - constrained applications, with small models, solving revenue producing problems like “how does my customer respond to this incentive?” These do not need big data centers.
Entirely new families of compute process device, signal, meter data from moving entities like autos, missiles, which require instant decisioning. The latency of sending info to a data center is architecturally fatal. Such compute systems cannot by definition operate with a centralized data center.
Again, Oracle, Palantir and other legacy big data center vendors are scamming the public, the press - elected officials that’s you - that the future is massive energy-consuming data centers.
They are public companies and their valuations depend on people who do not know better buying their stock - for their over-40 year old technology.
Remember, these tech dinosaurs use a tech tech stack Gartner - the undisputed authority on tech futures, said is obsolete in 2025. See BlackSwan on Substack for these facts and more.
the china framing is cope for the political goal, sure. but china is not having this fight at all. state coordination just routes around it. bytedance and alibaba get the land, the grid hookup, the local approval in the same planning cycle. the real gap is not the political problem. its whether the US can build at the same pace even after solving it.
Great post Jordan. UBI is a great idea - a variant is used for wind turbines in Aus (it usually a land fee, not strictly UBI). 10k might not move the needle for a lot of ppl. An industrial hum at night is pretty miserable. It will drive down home prices, rents and reduce quality of life. It will affect health and the IQ of kids that grow up in that area. I wouldn’t accept 10k a year to have a lower quality of life. That is a terrible deal. No amount of local amenity can make up for it. If I was a home owner I would anticipate house price drops of more than that. Do you pay the owner or the renter or both?
I totally agree with this. In my on the ground reporting in four different communities across the US over the past six months (Arizona, Texas, Michigan and Louisiana) I have been so surprised that AI data center developers don't consider throwing more money -- maybe a LOT more money -- at their problem of getting towns on board with their projects. treat these communities as the valuable resources they are, not a group to take advantage of with legal wrangling. It may not always work, but I think there would be a lot more support if the share were more valuable and fair.
I broadly respect the power of political pork, but (A) you'd need to be careful to not fall afoul of vote-buying laws with direct cash (B) village bribes may not circumvent larger state-level movements (C) presumably labs are cautious about starting a race to hell on welfare when they have S-1s to keep clean.
Maybe pinning the blame of anti-datacenter discourse on China will backfire. Or maybe it will be the only practical political lever remaining in 2027 to keep datacenters around. Exponential growth applies to populist movements too.
The diffuse-gains, concentrated-pain split is the right lever, but the concentrated cost is more concrete than general dislike. PJM's capacity costs jumped about 174 percent for the 2025-26 delivery year, and its own market monitor named data center load as the primary driver, which shows up directly in nearby residents' bills. A revenue-share check only flips a town if it clears that rate increase and stays visible.
Oracle, Palantir and other legacy big data center vendors are scamming the public, the press, that’s you - elected officials - that the future is massive energy-consuming data centers.
Any application requiring a data center, can be built with low I/O technology, and run on one or more Apple minis or Intel tiny computers, using less power than a Tesla, no water cooling, at 1/10th the cost and will run 100 to 1,000 to 1 million times faster - no data center needed.
Quantum computing companies are soon to be a dominant force in high compute problem solving. Trillions of dollars are being invested in quantum computing. Quantum computing does has no need for current data centers - none.
New technologies like virtual MESH are being demonstrated today, where the contents of an entire data center can run, instantly, on existing processors from a phone, tablet, server, Raspberry Pi - duplicating a data center with virtual no additional energy - or water - or environmental devastation.
A.I. applications are moving from LLM-centric “capture all knowledge” to where the money is - constrained applications, with small models, solving revenue producing problems like “how does my customer respond to this incentive?” These do not need big data centers.
Entirely new families of compute process device, signal, meter data from moving entities like autos, missiles, which require instant decisioning. The latency of sending info to a data center is architecturally fatal. Such compute systems cannot by definition operate with a centralized data center.
Again, Oracle, Palantir and other legacy big data center vendors are scamming the public, the press - elected officials that’s you - that the future is massive energy-consuming data centers.
They are public companies and their valuations depend on people who do not know better buying their stock - for their over-40 year old technology.
Remember, these tech dinosaurs use a tech tech stack Gartner - the undisputed authority on tech futures, said is obsolete in 2025. See BlackSwan on Substack for these facts and more.
the china framing is cope for the political goal, sure. but china is not having this fight at all. state coordination just routes around it. bytedance and alibaba get the land, the grid hookup, the local approval in the same planning cycle. the real gap is not the political problem. its whether the US can build at the same pace even after solving it.
Invite Erin Brockovich (or Julia Roberts) to the pod. She's on the case now: https://www.brockovichdatacenter.com/index.html
“The ladder he saws while climbing” lolol
NIMBYism is way more powerful than mere cash.
That said, getting some local projects done would be nice, of course
Great post Jordan. UBI is a great idea - a variant is used for wind turbines in Aus (it usually a land fee, not strictly UBI). 10k might not move the needle for a lot of ppl. An industrial hum at night is pretty miserable. It will drive down home prices, rents and reduce quality of life. It will affect health and the IQ of kids that grow up in that area. I wouldn’t accept 10k a year to have a lower quality of life. That is a terrible deal. No amount of local amenity can make up for it. If I was a home owner I would anticipate house price drops of more than that. Do you pay the owner or the renter or both?
Sure—but there has to be some dollar amount that can be spent to silence the datacenters!