The US is currently experiencing a rapid datacenter buildout but also a lot of anti-datacenter sentiment. Will the buildout be slowed substantially by mid-2029 for political reasons, compared to 2026?
I plan to resolve this question primarily by giving the resolution text to 2-3 frontier LLMs and asking them to resolve it. If the LLMs disagree, I'll use my personal judgment to resolve it, with a bias towards N/A.
Ways this question might resolve YES:
There's a federal ban on datacenters in the US
Most US states have passed datacenter moratoria, and the remaining ones aren't suitable for datacenters, such that few new hyperscale datacenters are built due to this
The US hasn't banned datacenters, but building datacenters in the US has become impractical or much more expensive, e.g. because the permitting for energy production has become much harder
Due to regulatory issues, the US has clearly become less attractive for building new hyperscale datacenters than other countries like Australia
Ways this question might resolve NO:
The current situation essentially continues
Nobody wants to build more datacenters due to an economic slowdown, but it would in theory still be as easy as before if the economic incentives were in place
AI training and inference are slowed/disincentivized, which reduces the economic incentive to build datacenters, but building datacenters would still be as easy as in 2026
1,000People are also trading
Softbank is loaning at junk bond levels+, NVIDIA investment alone exceeded all investment from the CHIPS Act, and datacenter capex exceeded total capex on all office space in the spring, no way. This would immediately choke the global economy and cause a deep recession, economic suicide.
In town hall meetings across the country, the likes of nebius, oracle, aws are constantly getting their way with ultimately ineffectual local opposition. The only path I see is if xrisk becomes politically mainstream or if the agent resolving this has an unusually permissive definition of "impractical or much more expensive." It'll definitely become a bit more expensive (already has, with many states passing large-load electricity tariffs on data centers), but won't be a blocker.
You're taking a decent sized stake in your own market.
@Velaris Yes, but note the AI resolution actually makes it very hard for me to mess with this market
@WilliamGunn I disagree and plan to keep it as-is, but I'll update it if lots more people complain about it
this seems egregiously mispriced to me. there's broad momentum towards data center restrictions across the country on the state level. if that doesn't play out, by the time all of the new data centers planned have come online in 2029, the power grid will be under much heavier strain in a way that seems like it will make this resolve YES on its own, without even considering factors like the skyrocketing cost of power transformers
the main scenario where this resolves NO is if there's a bubble pop and stock market crash and the datacenter buildout stops by itself...and like, lol, does the average manifold trader really still believe in that?
@Jack1 Yes, this is just about building – but if they cost a lot more to run, then they're less likely to be built