Well I shared the thread including your thought with Grok, and it said this:
-----
Here’s my actual, unbiased take on the technical debate (no Elon worship, no green hype, just physics and engineering reality as of 2025).
The original Slashdot skeptic is repeating a criticism that was 100% correct for the classic 1970s–2000s “hot dry rock” (HDR) projects (Los Alamos, Rosemanowes, Soultz 2000s era, etc.). Those really were conduction-limited. You drilled two wells a few hundred meters apart into 200–250 C granite, fracked a small volume, and watched the reservoir cool in months because heat was only replenishing by slow conduction through intact rock ( 2–3 W/mK). The recoverable fraction was tiny, and the impedance kept rising. That’s why HDR basically died commercially for 20 years.
Superhot / supercritical geothermal (400–450 C target, aiming for >500 C) is genuinely trying to escape that trap, and the physics does allow it—if everything works. Here’s why the critique no longer fully applies:
Energy density jumps by ~10× once you cross the critical point (374 C, 221 bar).
A kilogram of supercritical water at 500 C and 300 bar carries roughly 2,800–3,000 kJ/kg of usable enthalpy, versus ~300 kJ/kg for 250 C liquid water steam conversion in conventional geothermal. That alone reduces the mass flow (and therefore the parasitic pumping power and cooling load on the rock) by an order of magnitude for the same electrical output.
Heat transfer switches from conduction-dominated to convection-dominated.
In a well-designed, high-permeability fracture network, the supercritical fluid is buoyant and mobile. You get natural or strongly enhanced convection cells that sweep heat from a much larger volume of rock than the old HDR experiments ever touched. Surface area is now measured in square kilometers instead of hundreds of square meters.
Heat recharge rate in volcanic settings is completely different.
At Newberry, the background heat flux near the magma body is 300–1,000 mW/m (vs. ~80 mW/m continental average). Over a 1 km reservoir volume that’s hundreds of megawatts of continuous heat inflow. If your fracture cloud is big enough and permeable enough, you can pull 50–200 MW thermal more or less indefinitely without dropping the average rock temperature very much. (Fervo’s Project Red in Nevada—conventional temp, but still EGS—has already demonstrated ~3.5 MW from one well pair with negligible cooling over 18 months, and they’re only at ~190 C.)
Real remaining risks that the skeptic would be right to highlight:
Can you actually create and maintain a large, high-permeability fracture network at 500+ C and 10–20 km depth (or even 4–6 km near volcanoes) without it short-circuiting or sealing up with mineral precipitation? We don’t have field proof yet at those conditions.
Materials: drill bits, casings, pumps, and well cements that survive 500–600 C for 20–30 years do not fully exist off the shelf today. They’re being developed (high-nickel alloys, graphene-infused cements, etc.), but it’s expensive and unproven at scale.
Induced seismicity near an active volcano is non-trivial. Newberry is a shield volcano with low seismicity risk compared to, say, a subduction-zone stratovolcano, but still.
Economics: even if the resource works, LCOE has to beat new combined-cycle gas (~$40/MWh) and new nuclear (~$70–90/MWh) in the 2030s. The IEA’s “8 % of global electricity by 2050” scenario assumes superhot gets down to ~$30–50/MWh. That’s plausible only if wells become cheaper (thanks oil&gas tech transfer) and capacity factors stay >90 %.
Bottom line
The old “rock has terrible thermal conductivity, therefore geothermal is dead” argument is no longer a universal show-stopper for superhot volcanic/near-magmatic systems with modern EGS techniques. The physics really does flip in your favor above ~400–450 C. But turning that into thousands of gigawatts is still an engineering and materials challenge, not a solved problem. Mazama, Fervo, Eavor, Quaise, Sage Geosystems, etc., all have to demonstrate multi-decade performance at scale. The next 5–10 years of pilots will tell us whether the skeptics were merely early, or fundamentally right.
So: the ChatGPT answer you got was essentially correct on the physics and the generational shift. The Slashdot commenter is stuck in 1995 thinking, which is understandable—most of the industry was too until about 2018.
-----
If you think that this is incorrect information, a hallucination, or unduly sycophantic, I would like to hear your genuine case.