Alibaba’s Elements Claw AI agent unearths 4 new superconductors
Damo Academy unveils an AI agent able to discover superconductors, which could revolutionise scientific materials research and innovation
Alibaba Group Holding’s Damo Academy has unveiled what it calls the industry’s first artificial intelligence agent for discovering superconducting materials, saying that the tool has already found four previously unknown compounds that were later verified in laboratory experiments.
Discovering new superconductors has long relied on laborious, trial-and-error experiments because scientists still lack a complete theoretical framework to predict superconductivity. Over the decades, researchers have only accumulated about 2,000 known superconducting materials in the widely used SuperCon database.
The system was developed in collaboration with Renmin University of China and the University of Chinese Academy of Sciences.
Powered by a specialised, one-billion-parameter foundation model trained on 125 million molecular and crystal structures, Elements Claw screened 2.4 million stable crystal structures in 28 hours of graphics processor computing time.
It identified about 68,000 candidates with superconducting potential, which it then narrowed down to the most promising options for physical testing.
Rong Yu, head of scientific intelligence at Damo Academy, described them as the first superconducting materials discovered by an AI agent and subsequently confirmed in laboratory experiments, according to a post on the academy’s official WeChat account, although thousands of other candidates were yet to be explored.
Alibaba’s announcement comes as technology companies increasingly look beyond chatbots and coding assistants to apply AI to scientific research.
Instead of generating text or software code, these systems are designed to search scientific literature, analyse vast data sets and propose hypotheses that researchers can test in the laboratory.
Materials discovery is widely seen as one of AI’s most promising scientific applications because researchers must evaluate vast numbers of possible compounds before identifying a handful worth testing in the laboratory. AI could significantly shorten that process.
Huang Wenbing, an associate professor at Renmin University’s Gaoling School of Artificial Intelligence, said the framework for the Elements Claw could also be extended beyond superconductors to support the discovery of materials for solid-state batteries, catalysts and thermoelectric technologies.
Like Google DeepMind’s AlphaFold database, Alibaba is betting that opening its predictions to researchers could accelerate discoveries beyond what a single laboratory can achieve.
Alibaba owns the South China Morning Post.
Faster AI, lower costs: DSpark eases inference bottlenecks and chip strain, says DeepSeek
Start-up unveils speculative decoding framework that speeds up inference by up to 85 per cent amid China’s push to overcome US AI curbs
AI models’ conventional token-by-token output often slowed when responses were lengthy, leading to low utilisation of graphics processing units (GPU) and high user-perceived waiting time, which was a “primary bottleneck in serving AI”, the company said in research published on Saturday.
DeepSeek said the DSpark module accelerated AI response generation – also known as AI inference, which refers to serving a trained model to respond to user queries – by using a lightweight draft model to propose candidate responses and then verifying them in batches with a larger model, speeding up output.
DSpark further refined the approach with a semi-autoregressive generation method, allowing the model to produce small chunks of tokens rather than strictly one at a time.
It also introduced a confidence-based scheduling system that dynamically adjusted how much verification was applied based on computing demand, helping balance speed and output quality.
The new technique could reduce the computing resources needed to serve AI systems, according to Huang Yong, a Beijing-based programmer.
For example, with efficiency gains of up to 85 per cent, a single GPU that previously handled 100 user queries could now process about 185, he said.
DeepSeek tested the framework on several open-source models, including Google DeepMind’s Gemma and Alibaba Group Holding’s Qwen, suggesting DSpark’s enhancements could be applied broadly, with potential applications for companies seeking improved AI performance without significant computing resource investment. Alibaba owns the South China Morning Post.
The company has open-sourced its research of DSpark, a joint effort with the prestigious Peking University, on source code hosting platform GitHub and HuggingFace, the world’s largest online open-source AI community.
The release comes as Chinese AI developers face mounting pressure to make increasingly powerful models cheaper and faster to serve.
While Chinese AI models have been improving their general capabilities, the next battleground has shifted to AI inference optimisation, where companies seek to lower computing costs while handling surging demand from enterprise and consumer users.
The global AI boom has pushed up demand and prices for hardware infrastructure to serve these systems – from GPUs to memory chips – making efficiency gains critical.
DeepSeek’s introduction of DSpark followed comments by Shenzhen-based technology giant Tencent Holdings on Friday that inference efficiency has become a bottleneck to large-scale deployment of AI systems on inferior hardware.
It added it had made a series of engineering efforts – from attention mechanisms, asynchronous compute-communication to memory caching – to improve output speed.
Earlier this month, the AI team of smartphone-to-vehicle giant Xiaomi said it had improved the output speed for its MiMo-V2.5-Pro-UltraSpeed model to generate more than 1,000 tokens per second – one of the fastest output speeds in the industry.
US must innovate faster to counter China’s tech rise, lawmakers told
Witnesses urge investment in AI, robotics and chips, warning US research cuts risk eroding America’s technological edge
A US congressional hearing on Tuesday urged the United States to innovate faster, smarter and better to counter China’s growing technological muscle, even as several lawmakers slammed the US President Donald Trump administration for policies they said undercut US national interests.
“At this very moment, China has overtaken the United States in total R&D spending, while the administration has cancelled or frozen more than 7,800 research grants,” said Kathy Castor, a Democrat from Florida. “Even more concerning, our scientists are leaving our labs and increasingly out of the country.”
Lawmakers and witnesses called for passage of several key pieces of legislation aimed at addressing funding, support and guardrails for a range of technologies and supply chains from artificial intelligence and advanced robotics to quantum computing and semiconductors.
But passage of legislation is hardly a magic bullet if the US does not implement those laws and if it continues to undercut its own strengths, witnesses said.
Over 1,700 AI bills were introduced across America’s 50 states regulating AI, witnesses said, which can lead to a hodgepodge of inconsistent rules and incentives.
Kuaishou chip spin-off taps funding for in-house designs amid export controls
