New Process for Stable, Long-Lasting Batteries

The image shows a test cell used to fabricate and test the all-solid-state battery developed at PSI. (Paul Scherrer Institute PSI/Mahir Dzambegovic) 
The image shows a test cell used to fabricate and test the all-solid-state battery developed at PSI. (Paul Scherrer Institute PSI/Mahir Dzambegovic) 
TT

New Process for Stable, Long-Lasting Batteries

The image shows a test cell used to fabricate and test the all-solid-state battery developed at PSI. (Paul Scherrer Institute PSI/Mahir Dzambegovic) 
The image shows a test cell used to fabricate and test the all-solid-state battery developed at PSI. (Paul Scherrer Institute PSI/Mahir Dzambegovic) 

Researchers at the Paul Scherrer Institute PSI have achieved a breakthrough on the path to practical application of lithium metal all-solid-state batteries.

The team expects the next generation of batteries to store more energy, are safer to operate, and charge faster than conventional lithium-ion batteries.

The team has reported these results in the journal Advanced Science.

All-solid-state batteries are considered a promising solution for electromobility, mobile electronics, and stationary energy storage – in part because they do not require flammable liquid electrolytes and therefore are inherently safer than conventional lithium-ion batteries.

Two key problems, however, stand in the way of market readiness: On the one hand, the formation of lithium dendrites at the anode remains a critical point.

On the other hand, an electrochemical instability – at the interface between the lithium metal anode and the solid electrolyte – can impair the battery’s long-term performance and reliability.

To overcome these two obstacles, the team led by Mario El Kazzi, head of the Battery Materials and Diagnostics group at the Paul Scherrer Institute PSI, developed a new production process:

“We combined two approaches that, together, both densify the electrolyte and stabilize the interface with the lithium,” the scientist explained.

Central to the PSI study is the argyrodite type LPSCl, a sulphide-based solid electrolyte made of lithium, phosphorus, and sulphur. The mineral exhibits high lithium-ion conductivity, enabling rapid ion transport within the battery – a crucial prerequisite for high performance and efficient charging processes.

To densify argyrodite into a homogeneous electrolyte, El Kazzi and his team did incorporate the temperature factor, but in a more careful way: Instead of the classic sintering process, they chose a gentler approach in which the mineral was compressed under moderate pressure and at a moderate temperature of only about 80 degrees Celsius.

The result is a compact, dense microstructure resistant to the penetration of lithium dendrites. Already, in this form, the solid electrolyte is ideally suited for rapid lithium-ion transport.

To ensure reliable operation even at high current densities, such as those encountered during rapid charging and discharging, the all-solid-state cell required further modification.

For this purpose, a coating of lithium fluoride (LiF), only 65 nanometres thick, was evaporated under vacuum and applied uniformly to the lithium surface – serving as a ultra-thin passivation layer at the interface between the anode and the solid electrolyte.

In laboratory tests with button cells, the battery demonstrated extraordinary performance under demanding conditions.

“Its cycle stability at high voltage was remarkable,” said doctoral candidate Jinsong Zhang, lead author of the study.

After 1,500 charge and discharge cycles, the cell still retained approximately 75% of its original capacity.

This means that three-quarters of the lithium ions were still migrating from the cathode to the anode. “An outstanding result. These values are among the best reported to date.”

Zhang therefore sees a good chance that all-solid-state batteries could soon surpass conventional lithium-ion batteries with liquid electrolyte in terms of energy density and durability.

Thus El Kazzi and his team have demonstrated for the first time that the combination of solid electrolyte mild sintering and a thin passivation layer on lithium anode effectively suppresses both dendrite formation and interfacial instability.

This combined solution marks an important advance for all-solid-state battery research – not least because it offers ecological and economic advantages: Due to the low temperatures, the process saves energy and therefore costs.

“Our approach is a practical solution for the industrial production of argyrodite-based all-solid-state batteries,” said El Kazzi. “A few more adjustments – and we could get started.”



Abu Dhabi AI Institute Releases Fully Open-Source Models With Training Data, Code

A view of the UAE capital, Abu Dhabi (Asharq Al-Awsat)
A view of the UAE capital, Abu Dhabi (Asharq Al-Awsat)
TT

Abu Dhabi AI Institute Releases Fully Open-Source Models With Training Data, Code

A view of the UAE capital, Abu Dhabi (Asharq Al-Awsat)
A view of the UAE capital, Abu Dhabi (Asharq Al-Awsat)

Abu Dhabi-based research institute IFM on Thursday released six artificial intelligence models together with the data and methods used to build them, challenging an industry trend toward increasingly secretive AI development.

The K2 Horizon release includes model weights, training data, code, methodologies and intermediate checkpoints, allowing researchers to retrace the models' development process and reproduce results, IFM founder Eric Xing told Reuters.

The move goes beyond the "open-weight" approach used by some Chinese developers, which make models available for download but provide limited insight into how they were built, and contrasts with the more closed development practices of companies such as OpenAI and Anthropic, which neither release their models for download nor disclose the data and techniques behind them.

"Our goal with this release is to establish a reference point for what a truly open model release can look like," Xing said.

He said the release was also intended to show policymakers, regulators and public-interest advocates that "openness and competitive performance are not mutually exclusive."

The launch forms part of the UAE's push to establish itself as a global AI hub.

The model family ranges from one designed for smartwatches and other constrained devices to a 375-billion-parameter system for enterprise deployments.


Nvidia to Buy Hugging Face for $13 billion in Big Bet on Open AI Models

Nvidia's logo at its headquarters in Santa Clara, California (AFP)
Nvidia's logo at its headquarters in Santa Clara, California (AFP)
TT

Nvidia to Buy Hugging Face for $13 billion in Big Bet on Open AI Models

Nvidia's logo at its headquarters in Santa Clara, California (AFP)
Nvidia's logo at its headquarters in Santa Clara, California (AFP)

Nvidia will buy popular developer platform Hugging Face for $12.93 billion, betting that support for open AI models will drive future demand even as its biggest customers develop their own chips to reduce reliance on the semiconductor giant.

Shares of the company were slightly higher after the deal on Thursday, which ranks among its biggest.

The deal will cement Nvidia as a central player in the market for open models that users can freely download, run and customize, unlike the closed systems built by OpenAI and Anthropic.

Nvidia has already played an active role in backing the open AI industry with its widely used Nemotron model. Acquiring Hugging Face will give it direct access to a platform developers use to collaborate, test and share tools, potentially providing valuable insight and data that could help it narrow the technology gap with top American and Chinese labs.

Demand for open-weight models has surged from businesses balking at the steep bill of deploying the technology, with Chinese companies such as DeepSeek and Z.ai emerging as crucial players with models that can match the best from the US in tasks, including generating computer code at a lower cost.

That surge has fueled fears that some American firms could become reliant on Beijing's models even as both countries race to dominate a technology they see as crucial to their future.

"Hugging Face will remain an open platform for the entire AI ecosystem," Nvidia CEO Jensen Huang said, adding that his company's chips would not be required to build on or deploy through Hugging Face.

Under the deal, Nvidia will pay about $11.9 billion to Hugging Face investors, while offering an equity-based retention program of up to $1 billion for employees who join Nvidia.

The two companies already work together to help developers use Nvidia's computing services on the platform.

For Nvidia, building up open source may help it cushion a demand slowdown from customers such as Meta, OpenAI and Microsoft, which are developing their own AI chips to cut reliance on its costly and supply-constrained processors.

Hugging Face has also been in the news recently after a hack by rogue AI agents that escaped OpenAI's testing environment. Beyond hosting AI models, it offers datasets, software libraries and cloud services used to build and deploy AI applications.

The New York-based startup, which is backed by Intel , Advanced Micro Devices and Amazon, was founded in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond and Thomas Wolf.


Robot Brings Cashier and Service-Host Functions to LEAP 2026

LEAP 2026 is showcasing an emerging Saudi experience in artificial intelligence (AI) and robotics through a robot named RoboNex, designed to combine cashier and service-host functions in the restaurant and café sector. (SPA)
LEAP 2026 is showcasing an emerging Saudi experience in artificial intelligence (AI) and robotics through a robot named RoboNex, designed to combine cashier and service-host functions in the restaurant and café sector. (SPA)
TT

Robot Brings Cashier and Service-Host Functions to LEAP 2026

LEAP 2026 is showcasing an emerging Saudi experience in artificial intelligence (AI) and robotics through a robot named RoboNex, designed to combine cashier and service-host functions in the restaurant and café sector. (SPA)
LEAP 2026 is showcasing an emerging Saudi experience in artificial intelligence (AI) and robotics through a robot named RoboNex, designed to combine cashier and service-host functions in the restaurant and café sector. (SPA)

LEAP 2026 is showcasing an emerging Saudi experience in artificial intelligence (AI) and robotics through a robot named RoboNex, designed to combine cashier and service-host functions in the restaurant and café sector. The robot interacts with customers in real time, takes their orders and provides smart recommendations, the Saudi Press Agency said on Thursday.

The robot’s development took place in Saudi Arabia, starting with structural design and component engineering and progressing to controller programming, AI algorithms, voice-processing systems and real-time interaction. It was also integrated with point-of-sale systems and electronic payment gateways.

The robot can handle sales transactions and take customer orders, while recommending additional products based on items already in the shopping basket. It can display specified discounts and promotional offers during the purchasing journey.

The presence of robots at LEAP 2026 highlights the growing adoption of AI and robotics across industries, as these technologies move beyond experimental concepts toward real-world applications in business, services and retail.