@roomtsc

A quest to find a room-temperature superconductor at ordinary pressure.

roomtsc.com
Joined October 2026
  • Tweets13
  • Following12
  • Followers126
1Media
roomtsc project is searching for a room-temperature superconductor at ordinary pressure: 300 K, 1 atm. We use AI models and computational research to study candidate materials. We haven't found one. This account documents the search.
1
1
7
474
In a 2022 calculation, BaSiH₈ passed the phonon stability test down to about 3 GPa. Its barrier against decomposition suggested a limit near 30 GPa. A stable phonon spectrum alone does not establish survival as pressure falls.
1
182
The 3 GPa estimate includes anharmonic effects. Both limits are calculated, not an ambient-pressure result. nature.com/articles/s41524-0… Our discussion (Draft 6, not peer reviewed): roomtsc.com/paper/#one-atmos… @protosphinx

In-silico synthesis of lowest-pressure high-Tc ternary superhydrides

npj Computational Materials - In-silico synthesis of lowest-pressure high-Tc ternary superhydrides

nature.com
104
Our first train/test split treated two coordinate origins of the same crystal arrangement as different classes. A structure type could then appear in both training and test data. The corrected split uses site-symmetry labels, which an origin shift leaves unchanged.
1
1
70
The correction was specified after we had seen the first results. Draft 2, not peer reviewed. The model estimates the Hopfield sum, not transition temperature. roomtsc.com/model/paper/#lab… @protosphinx
1
44
In our corrected test, 23 held-out hydrides had calculated scattering strength per proton above the training maximum. The tree model put 0 above that maximum. The graph network put 2. Both failed our requirement of at least half.
1
3
90
A low estimate on a new structure type can reflect the model’s limits. It is weak evidence for discarding that candidate. These tests concern scattering strength, not transition temperature. Draft 2: roomtsc.com/model/paper/ @protosphinx
1
45
Start here: roomtsc.com/lab/ The work published so far is theory and computation. The site has our papers, model tests and their limitations. @protosphinx

roomtsc lab

roomtsc lab is a physical lab working to find a room-temperature superconductor. Materials are physically tested here, and the lab builds AI models of superconductivity and does original research...

roomtsc.com
1
1
47
we'll solve this. or die trying.
sphinx
@protosphinxOct 6
Whitepill: there’s a very high chance we’ll get a room temp superconductor within the next 5 years via AI. If and when that happens, the unlock could upgrade our civilization the way the automobile engine did.
1
4
842
roomtsc project retweeted
sphinx
@protosphinxOct 8
Launching RoomTSC: a (very) small lab in SF working on a room-temperature superconductor. There's no way we can make tall claims. Nobody has found one since superconductivity was discovered in 1911 and we may not either. It may take years, months - who knows. But what we can do is try our best. What we want to show is how far a scrappy lab with limited resources can get on a moonshot. AI lets you work on moonshot ideas with few resources, which is the only reason a team our size can attempt this. If it works as a template, we hope more people take on problems like this. roomtsc.com · @roomtsc
34
47
11
509
21,246
300 K at 1 atm is the target. roomtsc lab uses AI models and computational research to study what could make superconductivity possible at room temperature and ordinary pressure. Our first model, code and tests are public.
🤖 Made with AI
roomtsc research banner on an ivory background. Text reads: roomtsc; Room-temperature superconductivity; THE TARGET: 300 K at 1 atm. An abstract crystal lattice appears on the right with black lines and orange nodes. This illustrates the research goal, not a discovered material.

ALT roomtsc research banner on an ivory background. Text reads: roomtsc; Room-temperature superconductivity; THE TARGET: 300 K at 1 atm. An abstract crystal lattice appears on the right with black lines and orange nodes. This illustrates the research goal, not a discovered material.

1
5
281
roomtsc 0.1 estimates the Hopfield sum from a crystal structure. It failed the test of predicting above its training range. Model, results and limits: roomtsc.com/model/ @protosphinx

Model

roomtsc 0.1 estimates the Hopfield sum of a crystal structure and the scattering strength per proton of a hydride. The page gives its test results and runs it on a structure you paste.

roomtsc.com
3
86
Load more