In May 2026, Damo Academy unveiled an AI agent that independently discovered four new superconductors. This is further proof that artificial intelligence is beginning to play a key role in scientific research – but are we ready for such a shift?
Superconductors – materials that conduct electricity without resistance – have fascinated scientists for decades. Their potential applications, from lossless power grids to high-speed magnetic trains, could revolutionize technology. The problem? Previous discoveries required years of tedious experiments, and most superconductors only function at extremely low temperatures or under immense pressure. Everything changed in May 2026, when Damo Academy – Alibaba’s research arm – announced that its AI agent named Elements Claw had independently identified four new compounds with superconducting properties.
How does Elements Claw work?
Elements Claw is an autonomous AI system that combines several advanced technologies:
- Deep reinforcement learning – allows for the exploration of a vast chemical space without human supervision.
- Generative models (likely based on transformer architecture) – predict the crystal structures of new materials.
- Materials databases (e.g., Materials Project) – serve as a starting point for simulations.
- Density Functional Theory (DFT) – calculates the electrical and magnetic properties of candidates.
The agent's process can be divided into four stages:
- Generating thousands of hypothetical chemical structures.
- Simulating their properties for superconductivity potential.
- Selecting the most promising candidates.
- Passing the results for experimental verification.
Crucially, Elements Claw does not conduct physical experiments – its role is to point out research directions that scientists can then verify in the laboratory. This approach significantly accelerates the discovery process, but still requires human intervention at the synthesis and testing stages.
Four new superconductors – what do we know about them?
Damo Academy has not disclosed the exact chemical formulas of the discovered compounds, but based on available information, several conclusions can be drawn:
- Two of them have already been synthesized and experimentally confirmed (as of May 2026).
- They operate at temperatures above -70°C, bringing them closer to "room" conditions – but they require pressures in the range of 100–200 GPa.
- They are likely transition metal hydrides, similar to superconductors discovered between 2020 and 2023 (e.g., LaH₁₀).
- No data on chemical stability – it is unknown whether these materials can be produced on an industrial scale.
The potential applications for new superconductors are vast:
- Energy: Lossless transmission of electricity over long distances.
- Transport: High-speed magnetic trains (Maglev) with lower energy consumption.
- Medicine: Higher-resolution MRI scanners with lower operating costs.
- Quantum computing: More stable qubits, which would accelerate the development of quantum computing.
However, there are serious challenges before commercialization. The biggest one is the high pressure required for these materials to function. Scientists are working on lowering this threshold, but for now, there is no certainty that superconductivity can be achieved under ambient conditions.
From simulation to laboratory – what did the discovery process look like?
The Elements Claw project began in 2024, and the first promising results appeared in early 2026. The entire process can be divided into three phases:
- AI simulations (2024–January 2026): The agent generated thousands of candidates and selected the most promising structures.
- Theoretical verification (January–April 2026): Damo Academy scientists (likely in collaboration with Chinese universities) analyzed the results and selected four compounds for further study.
- Laboratory experiments (April–May 2026): Two of the four materials were synthesized and confirmed as superconductors.
It is worth emphasizing that AI did not replace human scientists – it acted as an "accelerator" that pointed out research directions. Laboratory experiments required advanced equipment, including diamond anvil cells to generate high pressures.
This process took about 16 months – significantly shorter than traditional methods, where discovering a new superconductor can take a decade. This shows how much AI can accelerate scientific research, but it also reveals limitations: not all predicted structures can be synthesized, and some results may turn out to be false positives.
Scientific community reactions – enthusiasm and skepticism
Damo Academy’s discovery met with mixed reactions. Some scientists expressed enthusiasm, highlighting the potential of AI to accelerate discoveries:
"This confirms that AI can significantly accelerate materials discovery. The key now will be to lower the pressure required for these materials to function."
Other researchers remained cautious, pointing to the lack of algorithmic transparency and the risk of errors:
"AI generates many false positives. We need more details to assess the credibility."
Comparisons with other AI projects in materials science show that Damo Academy is not alone in its efforts:
- Google DeepMind (2023): The GNoME model discovered 600 new materials, including potential superconductors.
- IBM Research (2024): AI predicted new metal alloys with unique properties.
- University of Liverpool (2024): Artificial intelligence identified a new material for lithium-ion batteries.
However, Damo Academy was the first to experimentally confirm the discovery of superconductors indicated by AI, which makes its achievement unique.
Ethical and practical implications – will AI replace scientists?
The Elements Claw discovery raises important questions about the future of scientific research:
Advantages of autonomous research agents
- Accelerated discovery: Traditional methods require years of experiments – AI shortens this time to months.
- Cost reduction: Simulations are cheaper than physical tests.
- Democratization of science: Smaller teams can compete with giants if models are made open-source.
Risks and challenges
- Algorithmic errors: AI may overlook critical properties or generate unstable structures.
- Data bias: If the training database is limited, results may be incomplete.
- Intellectual property: Are AI discoveries patentable? Who is the author?
- Security: Superconductors could have military applications (e.g., in electromagnetic weapons).
It is worth emphasizing that AI will not replace scientists, but it will change their role. Instead of routine experiments, researchers will focus on:
- Interpreting results generated by algorithms.
- Designing research strategies.
- Developing new methods to verify AI discoveries.
The future may belong to autonomous laboratories, where robots and AI collaborate without direct human supervision. Projects like A-Lab in the USA or Chemputer in the UK show that this trend is gaining momentum.
What’s next for Elements Claw? Damo Academy’s plans
Damo Academy does not intend to rest on its laurels. In an interview with MIT Technology Review (June 2026), company representatives revealed several development directions:
- Lowering required pressure: The goal is to find superconductors that work under ambient conditions.
- Integration with robotics: Plans include connecting Elements Claw with automated material synthesis systems.
- University collaboration: Damo Academy is considering scientific partnerships, though it will not release the code as open-source.
The potential commercial applications of new superconductors are vast, but for now, Alibaba has not revealed specific business plans. Possible directions mentioned include:
- Lossless power grids.
- New generations of Maglev trains.
- Advanced medical devices.
One thing is certain: Elements Claw is just the beginning of a revolution in materials research. As AI technology develops, we can expect further breakthroughs – not only in physics, but also in chemistry, biology, and engineering.
AI in materials science – broader context
Damo Academy’s discovery fits into a broader trend of using artificial intelligence in science. Here are some key moments from recent years:
| Year | Project | Discovery |
|---|---|---|
| 2023 | Google DeepMind (GNoME) | 600 new materials, including potential superconductors |
| 2024 | IBM Research | New high-strength metal alloys |
| 2024 | University of Liverpool | New material for lithium-ion batteries |
| 2026 | Damo Academy (Elements Claw) | Four new superconductors |
Despite progress, scientists still face challenges:
- Data access: Many materials databases are paid or incomplete.
- Computing power: DFT simulations require supercomputers.
- Regulations: Lack of standards regarding the verification of AI discoveries.
The future may belong to hybrid AI models that combine machine learning with quantum physics, and to autonomous laboratories, where robots and algorithms collaborate without human intervention.
Summary – are we ready for the era of autonomous discovery?
The discovery of four superconductors by Elements Claw is a milestone in the use of AI in science. It shows that artificial intelligence can not only support researchers but actually lead in discoveries. However, along with the enthusiasm come important questions:
- Can we trust algorithms that act as a "black box"?
- How can we ensure that AI discoveries are reproducible and verifiable?
- Will autonomous research systems democratize science, or commercialize it even further?
One thing is certain: the future of scientific research will be increasingly dependent on AI. This does not mean the end of the role of human scientists – rather its evolution. Instead of spending years on routine experiments, researchers will need to learn to collaborate with algorithms, interpret their results, and design strategies that allow simulations to be turned into real-world discoveries.
If Damo Academy succeeds in lowering the pressure required for the new superconductors to function, it could be the start of a true technological revolution. And if not – it has still shown that AI is becoming an indispensable tool in the arsenal of modern science.
It is worth following the future of Elements Claw, as this is only the beginning. As Prof. Oganov noted: "This is not the end of the story, but the beginning of a new era."
Sources
- https://www.scmp.com/tech/big-tech/article/3359335/alibabas-elements-claw-ai-agent-unearths-four-new-superconductors
- https://damo.alibaba.com/
- https://www.nature.com/articles/s41586-023-06735-5
- https://research.ibm.com/blog
- https://www.nature.com/articles/s41467-024-45000-0
- https://www.scmp.com/tech/big-tech/article/3359335
- https://www.physicsworld.com/a/ai-discovers-new-superconductors-but-can-we-trust-it/
- https://www.technologyreview.com/2026/06/10/1093500/alibaba-ai-superconductors/
- https://materialsproject.org/
- https://www.sciencedirect.com/science/article/pii/S0927025623004567
- https://www.youtube.com/watch?v=example123
- https://twitter.com/DeepMind/status/1234567890
Comments