Nvidia CEO Jensen Huang is in London, standing in front of a room full of journalists, outing himself as a huge fan of Gemini’s Nano Banana. “How could anyone not love Nano Banana? I mean Nano Banana, how good is that? Tell me it’s not true!” He addresses the room. No one responds. “Tell me it’s not true! It’s so good. I was just talking to Demis [Hassabis, CEO of DeepMind] yesterday and I said ‘How about that Nano Banana! How good is that?’”

It looks like lots of people agree with him: The popularity of the Nano Banana AI image generator—which launched in August and allows users to make precise edits to AI images while preserving the quality of faces, animals, or other objects in the background—has caused a 300 million image surge for Gemini in the first few days in September already, according to a post on X by Josh Woodward, VP of Google Labs and Google Gemini.

Huang, whose company was among a cohort of big US technology companies to announce investments into data centers, supercomputers, and AI research in the UK on Tuesday, is on a high. Speaking ahead of a white-tie event with UK prime minister Keir Starmer (where he plans to wear custom black leather tails), he’s boisterously optimistic about the future of AI in the UK, saying the country is “too humble” about the country’s potential for AI advancements.

He cites the UK’s pedigree in themes as wide as the industrial revolution, steam trains, DeepMind (now owned by Google), and university researchers, as well as other tangential skills. “No one fries food better than you do,” he quips. “Your tea is good. You’re great. Come on!”

Nvidia announced a $683 million equity investment in datacenter builder Nscale this week, a move that—alongside investments from OpenAI and Microsoft—has propelled the company to the epicenter of this AI push in the UK. Huang estimates that Nscale will generate more than $68 billion in revenues over six years. “I’ll go on record to say I’m the best thing that’s ever happened to him,” he says, referring to Nscale CEO Josh Payne.

“As AI services get deployed—I’m sure that all of you use it. I use it every day and it’s improved my learning, my thinking. It’s helped me access information, access knowledge a lot more efficiently. It helps me write, helps me think, it helps me formulate ideas. So my experience with AI is likely going to be everybody’s experience. I have the benefit of using all the AI—how good is that?”

The leather-jacket-wearing billionaire, who previously told WIRED that he uses AI agents in his personal life, has expanded on how he uses AI (that’s not Nano Banana) for most daily things, including his public speeches and research.

“I really like using an AI word processor because it remembers me and knows what I’m going to talk about. I could describe the different circumstance that I’m in and yet it still knows that I’m Jensen, just in a different circumstance,” Huang explains. “In that way it could reshape what I’m doing and be helpful. It’s a thinking partner, it’s truly terrific, and it saves me a ton of time. Frankly, I think the quality of work is better.”

His favorite one to use “depends on what I’m doing,” he says. “For something more technical I will use Gemini. If I’m doing something where it’s a bit more artistic I prefer Grok. If it’s very fast information access I prefer Perplexity—it does a really good job of presenting research to me. And for near everyday use I enjoy using ChatGPT,” Huang says.

“When I am doing something serious I will give the same prompt to all of them, and then I ask them to, because it’s research oriented, critique each other’s work. Then I take the best one.”

In the end though, all topics lead back to Nano Banana. “AI should be democratized for everyone. There should be no person who is left behind, it’s not sensible to me that someone should be left behind on electricity or the internet of the next level of technology,” he says.

“AI is the single greatest opportunity for us to close the technology divide,” says Huang. “This technology is so easy to use—who doesn’t know how to use Nano?”

Because lithium is relatively scarce and sodium is abundant in Earth’s crust, sodium-ion batteries are being investigated as viable, cost-effective alternatives to the widely used lithium-ion batteries. In these batteries, the choice of cathode material primarily influences battery capacity and stability.

Layered sodium manganese oxides (Na_2/3MnO₂) have attracted significant attention in recent years as cathode materials for high-capacity sodium-ion batteries without using any rare-earth metals. However, while these materials exhibit high initial capacity, their rapid capacity fading during charge-discharge cycling remains a significant challenge.

During charge-discharge cycling of NaMnO₂ electrodes, Na⁺ ions are constantly inserted and extracted from the cathode material. This is accompanied by changes in the oxidation states of manganese (Mn) between Mn³⁺ to Mn⁴⁺. When Mn³⁺ ions form, they distort their surrounding lattice to lower electronic energy, a phenomenon known as Jahn-Teller distortion.

Over time, these repeated distortions lead to a buildup of strain at both atomic and particle levels in NaMnO₂, eventually resulting in the loss of crystallinity and severe capacity degradation. This is the main cause of capacity loss during cycling of Na_2/3MnO₂ electrodes. Recent studies have attempted to address this issue by substituting metals at Mn sites.

In a recent study, a research team led by Professor Shinichi Komaba, along with Mr. Kodai Moriya and Project Scientist Dr. Shinichi Kumakura, from the Department of Applied Chemistry at Tokyo University of Science, Japan, revealed how scandium (Sc) doping can dramatically improve the cycling stability of P’2 polytype of Na_2/3MnO₂ electrodes.

“Previously, we discovered that Sc doping in P’2 Na_2/3[Mn_1-xSc_x]O₂ electrodes can improve the battery performance and long-term stability,” explains Prof. Komaba. “However, the exact mechanism for this improvement remains unresolved, and it was unclear whether this effect is generally applicable. In this study, we systematically studied P2 and P’2 polytypes of Na_2/3[Mn_1-xSc_x]O₂ to understand the role of Sc doping.”

Their study was published in the journal Advanced Materials on September 12, 2025.

The crystal structure of Na_2/3MnO₂ has several polytypes, which differ in several aspects. A key difference between the P2 and P’2 polytypes is that former exhibits localized Jahn-Teller distortions, while the latter features cooperative Jahn-Teller distortion where the distortions are aligned in a long-range order. The researchers conducted a series of experiments on both doped and undoped samples of each polytype containing varying amounts of Sc.

Structural tests revealed that Sc doping in P’2 Na_2/3[Mn_1-xSc_x]O₂ effectively modulates its structure, resulting in smaller particles and altered crystal growth, while preserving cooperative Jahn-Teller distortion and superstructure. This significantly improves structural stability. In addition, the team found that Sc doping prevents side reactions with liquid electrolytes and enhances moisture stability by forming a cathode-electrolyte interface layer.

As a result, in Na-half-cell tests, the Sc-doped P’2 type Na_2/3[Mn_1-xSc_x]O₂ electrodes demonstrated a substantial improvement in cycling stability. The sample with 8% Sc doping was found to have optimal performance.

The researchers also found that unlike non-doped samples, the crystallinity of the doped samples was remarkably maintained during cycling. Interestingly, Sc doping did not improve the cycling stability of P2 NaMnO₂ electrodes, indicating a specific synergy between Sc doping and cooperative Jahn-Teller distortion. Furthermore, doping with other similar metal cations, like ytterbium and aluminum, did not reduce capacity fading, highlighting the unique role of Sc.

They also tested the effect of pre-cycling, a common technique to improve cycle life, which further improved capacity retention in the doped P’2 Na_2/3[Mn_1-xSc_x]O₂ electrodes. Building upon these results, the researchers fabricated coin-type full cells using the 8% Sc-doped P’2 Na_2/3[Mn_1-xSc_x]O₂ electrodes, which demonstrated an impressive 60% capacity retention after 300 cycles.

“Since Sc is an expensive metal, our study demonstrates its feasibility in the development of batteries. Our findings can potentially lead to development of high-performance and long-life sodium-ion batteries,” says Prof. Komaba, highlighting the importance of their research.

“Moreover, beyond sodium-ion batteries, our study illustrates a new strategy to extend the structural stability of layered metal oxides involving the lattice distortion and improve the performance of batteries made using these materials.”

Overall, this study demonstrates the unique role of Sc doping for improving cycling stability of sodium-ion batteries, paving the way for their broader adoption.

In a world first, researchers at Nagoya University in Japan have successfully developed a resonant tunnel diode (RTD) that operates at room temperature made entirely from Group IV semiconductor materials.

The development of an RTD that operates at room temperature means the device could be deployed at scale for next-generation wireless communication systems. The use of only non-toxic Group IV semiconductor materials also supports more sustainable manufacturing processes.

This research marks a pivotal step toward terahertz wireless components that deliver unprecedented speed and data handling capacity with superior energy efficiency.

“Compared to InGaAs-based Group III-V RTDs that include toxic and rare elements, such as indium and arsenic, Group IV compounds-based RTDs are safer, lower cost, and offer advantages for creating integrated production processes,” said senior author Dr. Shigehisa Shibayama from the Nagoya University Graduate School of Engineering.

The results are published in the journal ACS Applied Electronic Materials.

Terahertz waves and quantum devices

Researchers have long struggled to achieve the high-speed and large-volume data transfer needed for sixth-generation (6G) cellular networks.

One promising solution is wireless communication using terahertz waves—electromagnetic waves that vibrate a trillion times per second, enabling ultra-high-speed data transmission. However, many technical challenges remain before this technology can be made practical for consumer applications.

A critical component for realizing terahertz communication is the RTD. This quantum device operates through negative differential resistance, a counterintuitive property where increasing voltage actually decreases current. When part of a properly designed circuit, this property allows the diodes to sustain high-frequency oscillations that would otherwise decay due to electrical losses.

Moving beyond laboratory constraints

The secret behind an RTD lies in its double-barrier structure, where electrons or holes tunnel through layers of different semiconductor materials, each only a few atoms thick. These layers have mainly been created from InGaAs-based Group III-V materials that include toxic and rare elements, such as indium and arsenic.

In previous research by the same group, the researchers created a p-type RTD using only Group IV materials, specifically germanium-tin (GeSn) and germanium-silicon-tin (GeSiSn) alloys. One limitation was that the diode only functioned at extremely low temperatures, around -263°C. Since consumer electronics and wireless systems cannot practically reach this level of cooling, the device would have remained a laboratory curiosity.

Shibayama and his colleagues have now discovered how to use only Group IV materials to produce a p-RTD that functions at room temperatures of around 27°C. This significant improvement opens new possibilities for the widespread adoption of terahertz semiconductor devices.

The research group achieved its breakthrough by introducing hydrogen gas during the layer formation process. They tested three different scenarios:

1. introducing hydrogen gas to both the two GeSiSn layers and three GeSn layers
2. introducing no hydrogen gas
3. introducing hydrogen gas to only the three GeSn layers.

In the last scenario, hydrogen gas restricted island growth and mixing between layers, resulting in a smooth and well-ordered double-barrier structure.

“The RTD cannot function if these layers are mixed,” said Dr. Shibayama.

“If there are defects in the layers, electrons can tunnel through these easier routes, leading to current leakage. This leakage current needs to be reduced for negative differential resistance—the key property of an RTD—to occur.”