As we move towards potential commercial viability or try and build some credibility in the space, it's important for us to set some goal posts and aim for them.
The discovery a room temperature superconductor is the ultimate target. To get there I think we need to balance our continued research on the topic with smaller, incremental milestones, that enable us to:
Build a brand around high tech discovery.
Foster new relationships and build credibility with leading researchers.
Attract investment and partnerships.
Milestone No. 1, Completion Target: May 2025:
A cloud platform that leverages MatterGen, our MLIP of choice, custom property predictors, and physics based DFT and MD simulation frameworks so any user can generate, and evaluate a .cif file. Each module would be completely independent so users could come to the platform with a .cif and simply run simulations.
This helps build a brand and begin to establish credibility. With an intuitive platform we can begin democratizing access to the SoTA in computational materials simulation and evaluation.
Monetization is enabled (and likely necessary to support the hosting costs), which starts to make a project like this attractive for private investment.
If someone in academia finds themselves on our platform and enjoys the experience it could open the door for new partnerships and collaboration opportunities.
Sub tasks could include:
Optimization of existing physics based models for scalable hosting.
Build out the website.
Develop property specific evaluation harnesses (ionic diffusion coefficient, magnetic moment, melting temp, etc).
Develop active learning loops for MatterGen.
Milestone No. 2, Completion Target: Fall 2025:
Published discovery of a novel and impactful material (my vote would be for an Iron based permanent magnet. Other options could include, solar cell improvements, carbon capture materials, cobalt free battery materials.). This discovery would be computationally driven from start to finish and could ideally be something that we could make in a garage without specialized equipment.
The impact here is hopefully obvious. Such an achievement puts us on the map.
Sub tasks could include:
Building partnership opportunities in academia and industry alike / networking within the field to better understand the SoTA that might not be published quite yet.
Modifications to MatterGen finetuning process (new property specific optimization functions?).
Outlining high level physical synthesis procedures that can be followed to go from crystal structure to bulk material.
Milestone No. 3, Completion Target: Early 2026:
Scaled simulation suites for bulk systems including defects and an expanded, much larger, diffusion model built on MatterGen. Riding the hype from milestone 2, the goal would be to thrust our methodology into the mainstream by leveling up the simulation possibilities and democratizing access to it. This gets a little bit back to the idea of providing a 'materials foundation model'. Only now we have a bit of a track record to lean on when publishing new results and methods.
At this point we could hopefully stand on results rather than hype. Milestone 2 proved that the process can work, and now everyone has the ability to try it themselves.
By now we're off to the races.
Room temperature superconductor research never stops:
Doing hard things just to do them will always be more valuable than we can ever realize in the moment. A continued focus on the room temp superconductors could very well make the discovery of materials with well understood behaviors trivial... And we want to be the folks to find them so why stop now.
Permanent magnet idea makes a lot of sense. There's a lot of parallels to room temperature superconductors so our research and simulations can borrow from one another. Both are emergent quantum mechanical effects, both effected by temperature, and of course both are crystals we can design and tune.
Really solid roadmap. Let's dive into it this afternoon. A few things coming to mind that I want to throw out:
My experience with self-service platforms has not been great. It's a really cool product to build but also really hard to get people to use it. Also, makes marketing hard IMO.
I don't think we know enough about our target audience yet. Would a self-service platform be useful to people in academia? I think so, but we should validate this and learn more about their current process.
I think it's really good to be opening our perspective to other kinds of materials and building repeatable pipelines that can help us optimize/discover those materials. That said, I don't see the difficultly being all that different for each material type (perm. magnet, solar cell, superconductor, etc.) Not that they're all easy, it's the opposite in that I think they're all equally hard. This could be very naive though. Actually synthesis is a whole different story.
I'm obsessed with superconductors because it's generally recognized as the hardest, most mysterious by the incumbent research community. Point above notwithstanding. We want to make noise in the community and I don't see us having the same impact, or any impact, by discovering other kinds of materials. Remember the hype of LK99? Let's do it for real. There's no other material that could generate that kind of buzz.