GE Aerospace and Magellan Aerospace have signed a memorandum of understanding to establish maintenance, repair, and overhaul capabilities in Canada for the F414-GE-39E engine that powers Saab’s JAS 39 Gripen E fighter.

The agreement, announced on July 22, would move forward only if the Canadian government selects the Gripen E for the Royal Canadian Air Force’s fighter fleet.

Under the proposed arrangement, GE Aerospace would train Magellan personnel and license the company as Canada’s domestic center of excellence for F414 engine sustainment. The work would be carried out at Magellan’s facility in Mississauga, Ontario.

Domestic engine support

“Today’s MOU signing builds on a six-decade relationship between GE Aerospace and Magellan that spans both military and commercial engines and will ensure the Royal Canadian Air Force has access to engine sustainment services in country to ensure readiness of the F414 engines,” said Paul Ferraro, Vice President of Defense Engines & Services at GE Aerospace.

The companies said the capability would support aircraft readiness, engine availability, and long-term fleet performance while giving Canada more control over fighter maintenance.

The proposed center would also extend Magellan’s role in military engine support by developing specialized infrastructure, technical expertise, and a skilled workforce within the country.

If implemented, the project would fit with Canada’s push to strengthen its domestic defense industrial base and retain more high-value aerospace work at home.

Sovereign aerospace capability

“This MOU reflects our shared intent to build enduring sovereign aerospace capabilities in Canada,” said Haydn Martin, Magellan’s Vice President in Business Development, Marketing, and Contracts.

“Should the Saab JAS 39 Gripen E aircraft be selected, Magellan Aerospace will be ready to provide world-class engine maintenance, repair and overhaul services that enhance operational readiness for the Royal Canadian Air Force while maintaining highly skilled Canadian jobs, developing advanced technical expertise, and strengthening Canada’s long-term defence industrial capacity,” he said.

For the Royal Canadian Air Force, having an in-country engine sustainment option could reduce dependence on foreign maintenance networks and improve access to repair capacity during periods of higher operational demand.

The planned capability would also support the development and retention of personnel trained to work on fighter propulsion systems.

F414 engine track record

The Gripen E uses the F414-GE-39E, part of GE Aerospace’s F414 engine family. According to GE Aerospace, the F414 has accumulated more than five million flight hours, while more than 1,600 engines have been delivered to eight nations.

The engine features a six-module architecture intended to simplify maintenance and allow components to be replaced or interchanged.

GE Aerospace also says the F414 operates without scheduled overhauls, relying instead on on-condition maintenance designed to maximize engine availability and reduce life-cycle costs.

The engine is built to provide rapid throttle response and unrestricted performance.

However, the Canadian sustainment plan remains tied to Ottawa’s fighter procurement decision. The proposed Mississauga operation would proceed only if Canada chooses the Gripen E, making the MOU both an industrial commitment and a potential element of Saab’s bid for Canada’s future air combat requirements.

The US Army is testing an experimental AI platform designed to dramatically accelerate battlefield planning by reducing decision-making timelines from days or hours to just seconds. The system, known as the Course of Action Generative Pre-trained Transformer (COA-GPT), is intended to help military personnel rapidly develop and adapt operational plans while maintaining human oversight throughout the process.

Developed by researchers at the US Army Combat Capabilities Development Command’s Army Research Laboratory, the Army’s primary fundamental research institution, COA-GPT is being evaluated as part of Project Convergence Capstone 6, the service’s largest experimentation exercise, taking place at Fort Irwin, California. 

The trial will assess how AI can support faster, more adaptable decision-making in future battlefield environments.

Tackling human-AI integration in military operations

The COA-GPT is an experimental AI platform built to speed up mission planning while providing commanders and intelligence teams with a more flexible framework for operational decision-making. 

The system is designed to address one of the military’s biggest challenges in adopting AI: ensuring effective collaboration between soldiers and AI. By combining AI-generated recommendations with human judgment, COA-GPT aims to help military personnel plan and adapt operations more quickly without removing people from the decision-making process.

According to Dr. Amar Marathe, program manager for COA-GPT at DEVCOM Army Research Laboratory, the platform combines generative AI with decades of military expertise and established doctrine to help Army leaders and soldiers generate, compare and evaluate different operational planning options. 

He stressed that the goal is not to replace human decision-makers, but to enhance their capabilities. Instead, COA-GPT is designed to fit into existing military planning workflows, enabling personnel to work faster, improve efficiency and make more informed decisions during mission planning.

Interface helps commanders guide autonomous systems

One of the key scientific challenges behind COA-GPT is enabling machines to better understand human intent across teams of soldiers and AI-enabled systems, according to Marathe. Through a partnership with Hyssos Tech, the platform incorporates an advanced graphical interface that allows leaders at different levels of command to communicate with AI more naturally. 

Using tools such as shared maps and visual inputs, commanders can better guide agentic AI systems while reducing miscommunication, cognitive burden and operational friction among personnel.

Marathe further explained that COA-GPT was developed with a flexible, plug-and-play architecture that allows researchers to replace individual components and experiment with new AI capabilities without rebuilding the entire system. This approach enables faster testing, adaptation and integration of emerging technologies as the platform continues to evolve.

Moving forward, the Army will continue refining the COA-GPT testbed through additional soldier feedback and collaboration with industry partners. Researchers will examine how generative AI could reshape the role of Army planning teams, identify the cognitive skills future planners will need and improve workflows to ensure the technology helps maintain a strategic advantage on the battlefield.

💾

New satellite imagery has been released that appears to show what analysts believe is a new kind of Chinese submarine. Captured by BlackSky back in May of this year, the new attack submarine appears to be under construction at the Jiangnan Shipyard in Shanghai.

Concrete details on it are thin on the ground, but analysts believe it could be an improved version of China’s existing Type 093 “Shang” class. If true, this would represent another step in China’s modernization of its undersea capabilities.

From what can be gathered, the new submarine appears to have a smaller conning tower than the Type 093, and appears to be shorter. The smaller tower could be significant as it could reduce drag and turbulence when submerged.

This would, in turn, reduce its acoustic signature, making it more stealthy. Hull design is also just one part of a vessel’s design. If the idea is to make it more acoustically stealthy, then the submarine would likely also feature other refinements like vibration isolation, anechoic coatings, sonar arrays, etc.

New Chinese submarine spotted

If true, this would make the submarine considerably harder to detect and track using things like passive sonar. For potential rivals in theatres like the Indo-Pacific, it could, in theory, thwart detection efforts from P-8 “Poseidon” aircraft, seabed sensor networks, etc. This would feed into China’s anti-access/area-denial strategy within its sphere of influence.

“Its role may be to silently approach U.S. aircraft carriers underwater and launch attacks,” former Vice Admiral Masao Kobayashi, who served as commander of the Japan Self-Defense Forces’ submarine fleet, told Asahi.

In reality, this translates to making it riskier for Western navies to operate near Taiwan or the South China Sea in general. If this new submarine is as quiet as believed, it could, in theory, stalk and potentially attack U.S. carrier strike groups.

Reduced drag would also translate into higher speeds when submerged, and much improved energy efficiency. That would enable it to travel further and operate for longer between resupply.

According to reports, the submarine appears to be around 393 feet (120 meters) long and 33 feet (10 meters) wide. It would also be just the latest in new submarines developed and launched by China in recent years.

Between 2021 and 2025, Asahi explains, China has deployed an estimated ten submarines, with a total displacement of around 79,000 tons between them. Beyond that, analysts are somewhat split on how to classify the new submarine.

New class or improvement?

Some believe it to be an entirely new class, while others prefer to see it as an improved variant of the Type 093. It is also not clear how the submarine is powered and propelled, with some believing it is likely nuclear-powered. However, it is impossible to tell from satellite imagery alone.

“The submarine appears to be a new class,” Timothy Heath, owner of research firm Perceptum, told the South China Morning Post (SCMP).

Whatever the case, it is important to note that we only have satellite imagery to go by right now. Chinese authorities have neither confirmed nor denied the new submarine’s existence.

💾

Despite fears that China could race ahead of the US in AI, a joint UK-US evaluation found that Moonshot AI’s Kimi K3 performs “significantly below” leading American models in offensive cybersecurity capabilities.

The findings address growing concerns in Washington regarding China’s recent advancements in artificial intelligence following the release of Kimi K3, described as China’s most powerful large language model.

The UK Artificial Intelligence Security Institute (UK AISI) and the US Center for AI Standards and Innovation (CAISI) found that Kimi K3 achieved an overall cyber capability score of 32.2%, compared to an average of 76.2% for top, unnamed US models.

Evaluation identifies cybersecurity gap

Researchers tested Kimi K3 using ExploitBench, a public benchmark developed by Carnegie Mellon University that measures an AI system’s ability to develop end-to-end exploits for software vulnerabilities.

According to the evaluation, Kimi K3 “performs significantly below the most recent frontier cyber-capable models” when tasked with developing exploits. The model attained a score of 32 percent, outperforming GLM-5.2, another Chinese open-weight model, which scored 24 percent.

However, Kimi K3 still lagged behind leading US systems. The most cyber-capable US models achieved an average score of 76.2%.

The report also examined whether models could achieve arbitrary code execution (ACE)—a high-severity exploit outcome that allows attackers to take control of a target system. Kimi K3 achieved ACE on none of the 41 ExploitBench tasks, while “the most cyber-capable models achieved ACE on 20/41 samples on average,” according to the UK AISI and CAISI.

Simulated attack shows remaining capabilities

The evaluation also tested Kimi K3 on “The Last Ones” (TLO), a 32-step simulated corporate network attack designed to measure a model’s ability to conduct end-to-end cyber operations autonomously.

Kimi K3 reached step 17 of the attack path on average, compared with the 28.5 steps achieved by the most cyber-capable US models.

Despite this gap, the report noted that Kimi K3 completed the simulated attack in one of 10 attempts. The UK AISI and CAISI said this demonstrated that the model “is capable of autonomously attacking small, weakly defended, and vulnerable enterprise systems, when directed to do so and given initial network access.”

Furthermore, Kimi K3’s safeguards did not prevent it from attempting cyber exploit development or offensive cyber operations during testing.

Although the results were only preliminary, the research noted that Kimi K3’s overall cyber capability score was estimated from a single benchmark, whereas other models were evaluated across a larger number of cyber tasks.

Findings reshape AI competition debate

Kimi K3 has sparked attention, if not concern, following reports of its strong performance on other AI benchmarks. However, these new findings offer a different picture of the model’s cybersecurity capabilities, showing that it still lags behind US systems.

The South China Morning Post reported that the evaluation “offers a stark reality check amid mounting panic in Washington over Beijing’s AI advancements following Moonshot AI’s latest flagship release.”

David Sacks, former White House AI czar and current co-chair of US President Donald Trump’s Council of Advisers on Science and Technology, said the findings suggest that fears over Chinese model supremacy may be exaggerated. He urged policymakers to avoid over-regulating the sector.

“The Kimi Panic needs to stop,” Sacks said in a social media post. “American frontier models are still ahead.”

💾

Air defense used to revolve around one basic problem: stopping enemy aircraft and missiles before they reached their targets. The rapid spread of drones has complicated that equation.

Cheap one-way attack drones, small reconnaissance UAVs, and larger strike platforms can now threaten military bases, infrastructure, and frontline units without requiring an adversary to risk an expensive crewed aircraft.

Traditional surface-to-air missiles, or SAMs, remain among the most powerful tools available for defending airspace. But using costly missiles against inexpensive drones can create an unfavorable exchange.

That has pushed militaries toward another option: drone interceptors designed specifically to hunt and destroy hostile UAVs. Here is how the two approaches compare across six key areas.

Cost per interception

The biggest difference is economics.

Modern surface-to-air missiles can be extremely expensive because they require sophisticated seekers, propulsion systems, guidance electronics, and warheads capable of destroying fast and highly maneuverable targets.

That makes sense when the incoming threat is a fighter jet, helicopter, or cruise missile. The calculation changes when the target is a relatively inexpensive drone.

Drone interceptors can potentially offer a lower-cost response. Many are built around simpler airframes, compact sensors, and reusable or comparatively inexpensive components. Some destroy targets by collision, while others carry small warheads or other defeat mechanisms.

The objective is simple: avoid spending a high-value interceptor on a low-value threat whenever possible.

Speed and target types

Surface-to-air missiles maintain a major advantage when speed matters.

SAM systems are designed to engage targets traveling at high subsonic, supersonic, and, in some cases, even greater speeds. That makes them suitable for threats such as combat aircraft and cruise missiles, depending on the system.

Drone interceptors generally operate against slower targets. They are particularly suited to small UAVs whose flight speeds and maneuverability allow another drone to pursue them.

Against a fast jet or missile, however, most interceptor drones would struggle to match the speed and engagement envelope of a dedicated SAM.

Range and coverage

SAM systems are built to defend large areas.

Long-range batteries can monitor and engage threats tens or even hundreds of kilometers away, depending on the weapon. This allows defenders to create layered protective bubbles around cities, bases, and strategic infrastructure.

Drone interceptors usually operate over much smaller areas. They may be launched once radars, electro-optical sensors, or other systems detect an incoming UAV.

Their shorter range makes them better suited to local defense rather than wide-area airspace control.

Ammunition and sustainability

Another important difference is how quickly each system can be replenished. A missile launcher carries a limited number of ready-to-fire interceptors. Once those missiles are used, the system must be reloaded, creating potential vulnerability during large or sustained attacks.

Drone interceptors could offer greater flexibility if they can be manufactured cheaply and stored in larger numbers.

That becomes particularly important during swarm attacks, where dozens of drones may arrive within a short period. The challenge is ensuring that interceptor drones can launch, coordinate, identify targets, and engage quickly enough to deal with multiple threats.

Guidance and autonomy

Surface-to-air missiles rely on highly mature guidance systems.

Depending on the weapon, they may use radar guidance, infrared seekers, command guidance, or combinations of several methods. Drone interceptors increasingly depend on autonomy.

Computer vision, artificial intelligence, onboard navigation, and target recognition can allow an interceptor to pursue another drone with limited human involvement.

That could make them useful against large numbers of small targets, but it also places significant importance on reliable detection, identification, communications, and autonomous decision-making.

Complementary weapons, not replacements

Drone interceptors are unlikely to make surface-to-air missiles obsolete. Instead, they are becoming another layer in increasingly complex air-defense networks.

SAMs remain essential when the threat is fast, long-range, heavily armed, or capable of operating at high altitude. Drone interceptors can fill the gap at the lower end of the threat spectrum, particularly against inexpensive UAVs that may not justify the use of an advanced missile.

The future of air defense will therefore probably not come down to choosing between missiles and interceptor drones. It will depend on using the right weapon against the right target.

As aerial threats become cheaper, more numerous, and more diverse, the most effective defenses will likely combine long-range missiles, guns, electronic warfare, directed-energy systems, and interceptor drones into a layered network capable of responding without exhausting its most expensive weapons.

The US Army and Air Force have signed 10-year Intergovernmental Support Agreements with Arizona State University to expand energy services across military installations in the continental United States.

Signed on July 7, the agreements allow installations to request support for utility monitoring, energy assessments, data integration, resilience planning, and workforce development.

The arrangement is intended to help bases strengthen energy resilience, improve infrastructure management, and prepare for operational disruptions.

Expanding access to energy expertise

Under the agreements, Army and Air Force installations can seek a broad range of technical services from Arizona State University.

These include utility metering and billing analysis, energy and water assessments, building automation, distribution network modernization, on-site generation, microgrid development, independent project reviews, energy readiness exercises, and workforce training.

“Intergovernmental Support Agreements give installations an important mechanism to access support services that can improve readiness and resilience,” said Jordan Gillis, assistant secretary of the Army for installations, energy and environment.

“This effort reflects the Army’s continued focus on strengthening infrastructure, improving efficiency and supporting mission requirements across our installations,” he continued.

The US Army said the partnership can also help installations better understand existing utility systems and identify where modernization or efficiency improvements are needed.

Managing aging utility systems

A major focus of the Army agreement is the condition and performance of utility infrastructure at military installations.

“Through this agreement, Arizona State University will provide the monitoring, analysis and technical support needed to help Army installations better understand and manage aging utility systems,” said David Dentino, deputy assistant secretary of the Army for installations, housing and partnerships.

“This partnership is about improving readiness, increasing resilience, reducing costs and ensuring our installations have the reliable infrastructure needed to support the mission,” he continued.

For the Air Force, the agreement similarly creates a route for installations to obtain services linked to infrastructure, energy efficiency, and mission assurance.

“National security requires energy and water surety,” said Nancy J. Balkus, deputy assistant secretary of the Air Force for infrastructure, energy and environment.

“Every one of our installations is not just a base; it is a power projection platform to assure lethality whenever called upon. Without energy and water security, our missions cannot test, launch, sustain or succeed,” he said.

Supporting long-term installation resilience

Intergovernmental Support Agreements allow military installations to obtain support services from eligible state and local government entities.

For the Army and Air Force, the new ASU agreements provide a framework through which bases can request energy-related expertise without creating separate arrangements for every service.

The broader goal is to improve the reliability of installation infrastructure while giving military facilities more options for monitoring, planning, training, and resilience projects.

By covering services ranging from metering and automation to microgrids and energy readiness exercises, the agreements give installations access to tools that can support infrastructure modernization and long-term operational preparedness.

The initiative also aligns with wider efforts to ensure military installations can continue meeting mission requirements as energy systems, infrastructure demands, and operating environments evolve.

US defense contractor Amentum has secured an $85 million contract to support the modernization of critical infrastructure within the US Navy’s Nuclear Propulsion Program (NNPP). 

The three-year agreement will focus on redesigning and upgrading spent fuel examination capabilities at the Naval Reactors Facility in Idaho, a key site responsible for supporting the Navy’s nuclear fleet. The new facility will allow scientists and engineers to analyze used fuel cores, comparing real-world performance data with predictions from advanced design and testing processes. 

The company points out that these efforts will help improve the safety, reliability, and long-term mission capability of the Navy’s nuclear-powered vessels.

Strengthening the Navy’s nuclear fuel evaluation capabilities

According to Mark Whitney, president of Amentum’s energy and environment business, the project reflects the company’s continued focus on providing the engineering expertise and operational support needed to sustain the US Navy’s nuclear-powered fleet. 

By upgrading critical spent fuel examination capabilities, Amentum aims to help the Navy better analyze, optimize, and extend the performance of nuclear fuel cores. The modernization effort is expected to provide engineers and scientists with improved tools and data to support future fleet readiness, safety, and reliability.

Whitney further noted that Amentum’s broader nuclear expertise positions the company to support some of the most complex challenges in the sector. He said the company’s capabilities cover the full nuclear project lifecycle, delivering safe, innovative, and reliable solutions for customers worldwide. 

He also added that Amentum’s integrated global nuclear expertise enables the company to take on demanding projects, including efforts that support the US Navy in maintaining the effectiveness of its nuclear-powered fleet.

Nuclear propulsion program marks decades of innovation and reliability

For more than seven decades, the Navy’s NNPP has remained at the forefront of nuclear technology, developing advanced systems to power the Navy’s global fleet. The program has built a reputation for combining innovation, safety, reliability, and operational efficiency, enabling nuclear-powered vessels to perform critical missions around the world. 

Through continuous research, engineering advancements, and rigorous testing, the NNPP continues to support the long-term performance of the US Navy’s nuclear capabilities, Amentum highlights. A collaboration between the Department of Energy and the Department of the Navy, the NNPP is responsible for overseeing the design, development, and operational support required to sustain the Navy’s nuclear-powered fleet. 

The program plays a central role in maintaining US maritime advantage by ensuring warfighters have access to vessels with exceptional endurance, stealth, responsiveness, and advanced warfighting capabilities. Through its work across the nuclear propulsion lifecycle, the NNPP helps support the reliability of the Navy’s strategic naval assets.

Furthermore, as operator of the world’s largest nuclear-powered fleet, including 11 active aircraft carriers and more than 70 submarines, the Navy relies on programs such as the NNPP and partners like Amentum to support the safety, performance, and long-term sustainability of its nuclear propulsion systems.

💾

American Fusion has officially cleared a major regulatory hurdle, securing research and development certification from the Texas Department of State Health Services (DSHS) for all twelve registered models of its Texatron Fusion Engine platform. 

The development was shared by American Fusion’s Executive Chairman Brent Nelson in an interview, as reported by the company. 

This regulatory clearance applies to system hardware configurations across a wide power capacity spectrum, beginning with 500-kilowatt units and extending to large-scale architectures designed to output up to one gigawatt. With authorization completed, technical teams are preparing to start the next phase of experimental testing at Texas Tech University.

The primary technical objective of the upcoming experimental sequence is to verify nuclear fusion reactions inside the Texatron confinement hardware. Operational research teams plan to gather diagnostic measurements concerning plasma dynamics, specifically tracking core plasma temperatures, particle densities, and magnetic containment behaviors during operational cycles. 

“Nelson stated that the Company’s initial testing objectives include validating fusion reactions within the Texatron platform and collecting performance metrics related to plasma behavior and system operation,” said the company in a press release.

Maintaining conditions for sustained fusion reactions

Evaluating these physical parameters allows engineers to assess how efficiently the system achieves and maintains the conditions required for sustained fusion reactions. Following the initial reaction verification, testing will shift toward evaluating the pulse-generated electrical output produced by the engine, which serves as an intermediate milestone before attempting continuous electricity generation.

“Data generated during testing is expected to be independently documented, with findings potentially available for publication following review,” remarked the company.

Unlike conventional nuclear fission reactors that require steam turbines and produce long-lived radioactive waste, the Texatron system utilizes magnetic-confinement technology to generate electricity from high-energy plasma while significantly reducing radioactive byproducts. 

“Advanced fusion systems can convert energy directly into electricity, reducing reliance on steam turbines and minimizing energy losses,” noted American Fusion. “This enables higher efficiency, simpler infrastructure, and faster, more flexible power generation for modern grids.”

Current engineering efforts are distributed across several intermediate engine sizes, including 5-megawatt, 10-megawatt, and 20-megawatt configurations, along with preliminary engineering planning for a 100-megawatt unit.

Transportable and reduced radiation profile

The physical characteristics and direct electrical output of the engine dictate its target integration environments. Hyperscale artificial intelligence data centers increasingly utilize high-voltage direct current power infrastructure, meaning matching the generator’s direct DC output reduces energy conversion losses compared to traditional steam turbines, though power conditioning is still required to stabilize voltage.

The transportable and reduced radiation profile of the hardware allows smaller units to be deployed on trucks to supply power for defense installations, mining operations, and remote industrial facilities. 

The operational targets for 2026 depend on executing the multi-stage plasma validation tests, establishing performance metrics, expanding patent protections, and negotiating initial power placement agreements with commercial partners.

“American Fusion Energy is moving from proven laboratory breakthroughs toward scalable energy systems by focusing on real-world engineering challenges,” concluded the company.  

“Building on validated progress in high-temperature plasma confinement, the company is translating fusion physics into practical pathways for future commercial power.”

US firm Rescale, a digital engineering platform that combines AI, simulation, and high-performance computing to help R&D and engineering teams design, test, and bring new products to market faster, has received funding under the Genesis Mission, a national initiative led by the US Department of Energy to build an advanced integrated science discovery platform.

As part of the program, Rescale will join Lawrence Berkeley National Laboratory, Lawrence Livermore National Laboratory, and Oak Ridge National Laboratory as a funded partner in the Agentic HPC Pipeline Initiative (AHPI). 

The initiative aims to use agentic AI on Rescale’s platform to help accelerate the adoption of national laboratory-developed computing tools and simulations by US industry.

Doubling scientific research output with AI and advanced computing

The Genesis Mission is an ambitious initiative aimed at accelerating scientific and technological breakthroughs by combining artificial intelligence, supercomputing, quantum systems, and advanced scientific instruments into a single research platform. 

Established through an executive order in November 2025, the program brings together the department’s 17 national laboratories and around 40,000 scientists and engineers to advance research in areas including energy, scientific discovery, and national security. 

Department officials have compared the scale and ambition of the initiative to landmark US efforts such as the Manhattan Project and the Apollo program, with the goal of doubling the productivity of the country’s scientific research within the next decade.

The US Department of Energy launched the first Genesis Mission funding round earlier this year, awarding $293 million to projects focused on areas including advanced manufacturing, biotechnology, nuclear energy, and quantum science. The program received more than 8,000 applications, making it the department’s largest-ever funding call.

Rescale’s AHPI was selected for Phase I funding to address a key challenge: helping industry access advanced simulation tools developed by US national laboratories that have often required specialized expertise and infrastructure.

Helping manufacturers access advanced AI-powered simulations

The company will work with three US national laboratories to bring agentic AI capabilities to advanced simulation tools. The project will integrate AI agents into WarpX, LiDO, and Adamantine, codes developed by Lawrence Berkeley National Laboratory, Lawrence Livermore National Laboratory, and Oak Ridge National Laboratory.

Using the Rescale platform, AI agents will help engineers choose simulation tools, configure inputs, monitor runs, and analyze results. The initiative aims to reduce the time and specialized expertise needed to use these simulations by more than five times, making advanced research capabilities more accessible to US manufacturers.

According to Rescale CEO Joris Poort, the Genesis Mission highlights the importance of making industrial AI technologies available to leading research and development teams that can apply them in real-world applications.

Building on Rescale’s existing partnership with ORNL’s Manufacturing Demonstration Facility, AHPI addresses a common gap: DOE-industry collaborations run for a fixed term, and once they end, partners keep the insight but lose the tools to act on it. AHPI keeps the resulting agentic workflows, surrogate models, and code environments live on Rescale, letting labs and industry keep collaborating long after a project’s funding ends.

💾

The race to build larger and more powerful quantum computers has mostly focused on engineering: adding more qubits, reducing noise, and keeping fragile quantum states alive. However, a new theoretical study suggests that another, largely overlooked problem may emerge as these machines grow. 

Researchers at the Helmholtz-Zentrum Dresden-Rossendorf (HZDR) have found that a strange quantum phenomenon known as the quantum Zeno effect could cause certain quantum computers to slow dramatically—or in the worst case, nearly stop computing altogether.

“For quite generic conditions, we find that the quantum Zeno effect poses strong limitations on the performance” of quantum systems, the study authors note.

Rather than exposing a flaw in today’s devices, this research work identifies a fundamental scaling challenge that designers may have to confront as quantum computers become larger. 

In their new study, the HZDR team says the effect could become especially important for adiabatic quantum computers, a class of machines widely studied for solving optimization problems, searching massive datasets, and simulating molecules.

When the environment won’t stop “looking” at a quantum computer

The quantum Zeno effect is one of quantum mechanics’ most counterintuitive ideas. It predicts that repeatedly measuring a quantum system can prevent it from changing its state, effectively freezing its evolution. 

The concept has been studied experimentally for decades. Earlier research works explored this phenomenon mainly in isolated laboratory systems or as a way to deliberately control quantum states. For example, a theoretical work showed that continuous measurements can trap quantum systems within specific states. 

The HZDR researchers asked a different question—what if a quantum computer experiences this effect unintentionally?

Adiabatic quantum computers work differently from gate-based quantum computers. Their “algorithms are considered robust and can be executed by quantum computers largely independently of the hardware that is used,” Ralf Schützhold, one of the study authors and the director of the Institute of Theoretical Physics at HZDR, said.

Instead of performing a sequence of quantum logic operations, they gradually reshape an energy landscape while keeping qubits in their lowest-energy, or ground, state. If this transformation is slow enough, the ground state naturally evolves into the solution to the problem being solved. 

To investigate, they built a theoretical model of an adiabatic quantum computer interacting with a realistic environment. In practice, no quantum processor is perfectly isolated. Even inside refrigerators cooled to temperatures just above absolute zero (−273.15°C), tiny amounts of heat, electromagnetic radiation, or other external influences still interact with qubits. 

These interactions cause decoherence, gradually disrupting the delicate quantum properties such as superposition and entanglement—that give quantum computers their power. In effect, the surrounding environment behaves like an unintended detector that continually checks the qubits’ state, even though nobody is intentionally measuring them.

More qubits could make the problem worse

The team analyzed the adiabatic version of Grover’s search algorithm, a famous quantum algorithm that searches unsorted databases much faster than classical methods. 

In adiabatic computing, the algorithm starts with an easy-to-prepare quantum state and then slowly transforms the system’s governing energy landscape, known as the Hamiltonian, until the ground state represents the desired solution. If the transformation is sufficiently slow, the qubits naturally remain in the ground state throughout the computation.

Their calculations revealed that scaling creates an unexpected vulnerability. As the problem size grows, the minimum energy gap separating the ground state from higher-energy states can become increasingly small. 

As these energy differences shrink, the computation must proceed more slowly to remain adiabatic. Since the computation takes longer while environmental disturbances continue acting at roughly the same rate, the quantum Zeno effect becomes increasingly likely. 

“Each disturbance acts like an unwanted measurement, slowing down the system’s evolution. In the worst case, a calculation could even freeze completely,” Gernot Schaller, one of the study authors and the head of Quantum Technologies at HZDR’s Institute of Theoretical Physics, said.

Much like a cake rises properly only if the oven remains closed, constantly opening the door to check on it interrupts the baking process, causing the cake to stop rising or even collapse. In the same way, repeated environmental disturbances keep interrupting the natural evolution of the quantum state before it can reach the desired endpoint.

More importantly, the study does not suggest that all quantum computers face the same limitation. The work focuses on adiabatic quantum algorithms and related approaches such as quantum annealing, where the computation relies on gradually crossing small energy gaps. 

A roadmap for building more resilient quantum computers

The study is theoretical rather than experimental, so the findings still need to be explored in real quantum hardware. 

However, the researchers say the problem can be mitigated by improving shielding against heat and electromagnetic interference and by using techniques such as spin echo, which applies carefully timed pulses to reduce interactions between qubits and their environment.

The findings also suggest that scaling quantum computers will require more than simply adding qubits. Future systems may need better protection from environmental noise and smarter control techniques to reduce the risk of the quantum Zeno effect becoming a bottleneck. 

The study is published in the New Journal of Physics.

A Chicago-based company has been awarded an initial $23.67 million acquisition contract under the U.S. Department of War’s Accelerate the Procurement and Fielding of Innovative Technologies (APFIT).

With this Numat Technologies will accelerate the deployment of next-generation chemical and biological protection suits for U.S. Special Operations Forces.

The award is being executed in partnership with the Defense Threat Reduction Agency (DTRA) and is intended to speed the transition of the company’s technology from development into operational use.

Numat will deliver field-ready protective suits at scale

The APFIT program was established to bridge the gap between promising defense technologies and field deployment by providing procurement funding for mature innovations that are ready for operational use. Since its launch, the program has supported the rapid acquisition of advanced capabilities for the U.S. military, with total awards now exceeding $2 billion, according to a press release.

“Over the past 10 years, Numat built the MOF industrial base for this moment, where we can activate an entire supply chain, from material design through high-volume system manufacturing, to protect the Warfighter,” said Ben Hernandez, CEO of Numat Technologies.

“This contract award reflects our long-term investment in American innovation. Numat will deliver finished, field-ready protective suits at scale, made completely in the United States.”

Numat’s protection suits incorporate metal-organic frameworks

Numat’s protection suits incorporate metal-organic frameworks (MOFs)—highly porous engineered materials designed to capture and neutralize hazardous chemical and biological agents. Unlike conventional protective garments that primarily act as barriers, the company’s MOF-enabled suits are engineered to self-detoxify by neutralizing harmful substances upon contact, while also improving breathability and wearer mobility.

Under the APFIT contract, Numat will supply the MOF-enabled chemical and biological (CB) suits directly to U.S. Special Operations Forces, whose personnel often operate in environments where exposure to chemical and biological threats is a significant concern. According to the company, the suits are designed to offer enhanced protection without compromising comfort or operational effectiveness.

The contract also represents a milestone for the domestic manufacturing of metal-organic frameworks, a technology that has gained increasing attention for applications ranging from defense and industrial filtration to energy storage and environmental protection. Numat said it has spent more than a decade building a U.S.-based supply chain capable of designing, producing and integrating MOFs into mission-critical systems.

Company Chief Executive Officer Ben Hernandez said the award reflects years of investment in establishing a domestic industrial base for MOF technology and positions the company to deliver field-ready protective equipment manufactured entirely in the United States.

The latest award builds on Numat’s long-standing collaboration with several U.S. defense organizations, including the Defense Threat Reduction Agency, the Office of the Assistant Secretary of War for Industrial Base Policy, the Capability Program Executive Office for Chemical, Biological, Radiological and Nuclear Defense, U.S. Special Operations Command, the U.S. Army and the U.S. Navy. These partnerships have helped advance MOF technology from laboratory research to scalable manufacturing, according to the company.

Beyond military applications, Numat said its MOF platform has potential uses across industrial, semiconductor and energy sectors, where advanced materials are increasingly being deployed for gas storage, chemical separation and hazardous material filtration.

The APFIT award underscores the U.S. government’s broader efforts to accelerate the procurement of innovative defense technologies while strengthening the domestic industrial base and reducing the time required to transition emerging capabilities into operational service.

Albany Engineered Composites and A&P Technology are combining advanced braiding and resin transfer molding technologies to accelerate the development of scalable composite solutions for future aerospace and defense applications.

Albany Engineered Composites (AEC), a segment of Albany International Corp., and A&P Technology, a global leader in precision braided composite reinforcement technologies, announced a strategic collaboration to explore and develop advanced braided composite manufacturing solutions for current and next-generation aerospace and defense applications.

The collaboration will capitalize on A&P Technology’s expertise in braided composite architectures and Albany Engineered Composites’ advanced Resin Transfer Molding (RTM) manufacturing capabilities. The companies said the partnership is focused on accelerating the industrialization of lightweight, highly integrated composite structures designed to support future commercial and defense aircraft production requirements.

“Future aerospace platforms demand manufacturing technologies capable of simultaneously improving performance, reducing recurring cost and scaling to production rates the industry has not previously achieved,” said Chris Stone, president of Albany Engineered Composites in a press release.

“By combining A&P Technology’s deep expertise in advanced braiding with Albany’s industry-leading RTM capabilities, we are creating a force multiplier for our customers, accelerating the transition from innovative textile architectures to repeatable, production-ready composite solutions.”

The companies will work together on research, technology development, manufacturing process maturation, and commercialization opportunities focused on purpose-designed braided composite structures. These efforts will target current and next-generation aeroengine platforms, commercial aircraft, Advanced Air Mobility (AAM) blades and structures, Collaborative Combat Aircraft (CCA), and solid rocket motor nozzles.

Expanding advanced aerospace capabilities

Albany Engineered Composites is a rapidly growing division of Albany International Corp. that designs and manufactures advanced composite components for aerospace and defense industries. Headquartered alongside its parent company in Rochester, New Hampshire, AEC operates advanced research, development and manufacturing facilities across the globe, including major composite production sites in Salt Lake City, Utah; Boerne, Texas; and Commercy, France.

As aerospace manufacturers seek larger, more integrated composite structures, the companies said that combining braided composite preforms with automated RTM processing provides a pathway to scalable manufacturing by reducing assembly complexity and supporting higher production rates.

“It’s a privilege to collaborate with Albany given their broad and deep design and manufacturing skill and know-how built on decades of success making parts for the most demanding applications,” said Andy Head, president of A&P Technology.

“Together, we can offer composite solutions not only at an enabling rate but also with enabling cost and performance benefits. We look forward to meeting customer challenges together.”

Brent Stevenson, vice president of emerging markets and technology at Albany Engineered Composites, said that customers are seeking integrated solutions that combine multiple manufacturing processes.

“Our customers are looking beyond individual manufacturing processes; they’re looking for integrated solutions,” said Stevenson.

“Braiding and RTM are naturally complementary technologies. Together, they enable highly engineered composite architectures supporting complex geometries, efficient manufacturing and production scalability for both current and next-generation aerospace and defense platforms.”

Supporting future aerospace platforms

Through the collaboration, Albany Engineered Composites and A&P Technology aim to advance composite manufacturing solutions that combine precision braided reinforcement technologies with advanced RTM processing. The companies’ work will focus on production-ready composite structures for a range of aerospace and defense applications, including commercial aircraft, military platforms and emerging aviation technologies.

The partnership brings together two specialized manufacturing approaches as the companies pursue opportunities in research, development, process advancement and commercialization of advanced composite solutions, the press release concludes.

💾

As AI models continue to grow in size and complexity, chipmakers are shifting their focus from individual processor performance to the ability to efficiently connect thousands of accelerators into a single computing system. 

These highly interconnected server architectures have become a key competitive area for AI infrastructure companies, which are racing to deliver the computing scale needed for models with trillions of parameters and the growing adoption of AI agents.

According to Ding Yunfan, vice president of AI framework architecture at Biren, improvements in the performance of a single graphics processing unit (GPU) are no longer enough to meet these demands. Instead, the industry’s biggest challenge is turning vast GPU clusters into unified, seamless computing platforms capable of operating as a single system.

Architecture designed to break GPU cluster limits

Biren believes the future of large-scale AI computing depends on replacing conventional electrical interconnects with optical technology. According to Ding, traditional copper-based connections are approaching their physical limits, effectively capping today’s server architectures at around 128 GPUs, the South China Morning Post reports.

To overcome that bottleneck, the company has developed a distributed, decoupled supernode architecture built around near-packaged optics (NPO), which positions optical fibres closer to the chips to increase bandwidth and improve data transmission. Biren said the approach could enable clusters with up to 1,024 AI accelerator cards, significantly expanding the scale of computing systems needed to support increasingly demanding AI workloads.

Ding said optical interconnects are becoming essential for overcoming the scalability limits of conventional server architectures. Shanghai-based Biren is among a growing number of Chinese companies tackling these infrastructure challenges. Rivals including MetaX, Enflame and Alibaba are also developing supernode architectures to compete with Nvidia’s NVL72 and next-generation NVL144 AI platforms.

Biren’s Shanghai-based rival MetaX has introduced its Xijing S600 AI supernode, featuring a single rack that connects 64 GPUs. The company said its design reduces signal loss by removing external cabling between compute nodes, switch nodes and interconnect systems, improving efficiency for large-scale AI workloads.

Looking beyond electrical links to build larger AI supernodes

Biren is also working on new approaches to expand the scale of its AI computing systems. The company’s orthogonal hardware architecture, co-developed with telecommunications giant ZTE, uses traditional electrical connections, while Biren is simultaneously testing a prototype near-packaged optics (NPO) interconnect system. 

The chipmaker said the optical approach could eventually support supernodes with more than 512 accelerator cards, helping address the growing demand for larger and more efficient AI infrastructure.

Despite growing interest in optical networking for AI infrastructure, Ding said the semiconductor industry is still in the early stages of developing and testing optical interconnect technologies. He noted that widespread adoption will require further advances before these systems can be deployed at scale.

Furthermore, Ding emphasized that the technology still requires real-world validation before it can be widely adopted. He projected that NPO optical systems could reach mass commercial deployment around 2028, as the industry continues refining the technology and addressing remaining engineering challenges.

💾

Austin-based SuperCritical Materials Corp. has secured a license from the U.S. Department of Energy (DOE) to commercialize a patented uranium and critical materials extraction technology developed by DOE laboratories.

The development is a significant step toward expanding future nuclear fuel supply through unconventional uranium resources.

Licensed technology is designed to extract uranium from seawater

Developed with support from the DOE’s Office of Nuclear Energy and led by researchers at Pacific Northwest National Laboratory (PNNL), the licensed technology is designed to extract uranium from seawater using advanced adsorbent materials. SuperCritical said the agreement grants the company rights to manufacture and deploy the technology at industrial scale in the United States, with plans to expand into allied nations in the future.

The announcement comes as the United States accelerates investment in advanced nuclear energy to meet rising electricity demand driven by artificial intelligence, data centers, advanced manufacturing and other energy-intensive industries. Industry experts have increasingly highlighted that while reactor development is advancing, expanding the upstream nuclear fuel supply chain—including uranium production, enrichment and fuel fabrication—remains a critical challenge.

According to the company, the world’s oceans contain an estimated 4.5 billion metric tons of dissolved uranium, representing a resource far larger than known terrestrial reserves. While extracting uranium from seawater has long been considered technically feasible, commercial deployment has remained limited due to economic and engineering challenges.

SuperCritical intends to commercialize DOE-developed process

SuperCritical said it intends to commercialize the DOE-developed process using advances in adsorption chemistry and materials science. In addition to uranium, the company believes the technology could also enable the recovery of other strategically important critical materials from seawater.

The company describes its long-term strategy as building infrastructure to support what it calls the “Intelligence Economy,” a period of economic growth expected to be driven by artificial intelligence, robotics, advanced computing, defense technologies and digital infrastructure. It argues that reliable nuclear energy and secure fuel supplies will be essential to sustaining this growth.

“Our objective is straightforward,” said Alexander Canon Bryan, founder and chief executive officer of SuperCritical Materials. “If the Intelligence Economy requires abundant, reliable nuclear energy, then it will also require abundant, reliable nuclear fuel.”

SuperCritical noted that its technical team includes former researchers from Pacific Northwest National Laboratory, including Dr. Gary Gill, a co-founder of the company and a scientist who has spent more than two decades researching uranium extraction from seawater.

The company said commercial deployment of the technology could contribute to strengthening domestic nuclear fuel supply chains while supporting broader U.S. efforts to improve energy security, expand critical materials production and reduce dependence on imported strategic resources.

The licensing agreement also aligns with ongoing federal initiatives aimed at strengthening America’s nuclear industrial base and increasing domestic production of critical minerals and materials.

Although seawater uranium extraction has been demonstrated in research settings, analysts note that large-scale commercialization will depend on achieving competitive production costs and successfully scaling the technology for industrial use. Industry observers view the DOE licensing agreement as an important milestone toward evaluating the commercial viability of the process.

MIT researchers have developed a new framework that could help energy planners decide where to build renewable energy projects as climate change puts growing pressure on power grids.

The study combines fine-scale meteorology with detailed simulations of energy infrastructure. Researchers found that where new energy projects are built could significantly affect whether future power systems can meet demand reliably.

The team applied the framework to decarbonized energy systems in New England and Texas. Their analysis found that systems designed around historic climate conditions could experience up to a fivefold increase in energy shortfalls by 2050.

By incorporating future climate conditions into energy planning, however, the researchers found that both regions could improve grid resilience at little or no additional cost.

“As we mitigate climate change with renewables, we can also adapt to climate change by using future weather projections in our power system planning, and the extra costs of that adaptation are, at least in this study, not much,” said senior author Michael Howland, an MIT professor.

Climate change adds new risks to power grids

Electricity systems are already changing rapidly as demand rises from trends such as artificial intelligence and transportation electrification, while falling costs drive greater adoption of wind and solar power.

But renewable generation and electricity demand are both affected by weather, creating new challenges as climate conditions change.

“This study explores the joint, simultaneous impacts on multiple components of the energy system, similar to compound events studied in climate science,” Howland explained.

“An extreme weather event can impact wind and solar generation and electricity demand all at the same time. Our hypothesis is that’s likely to be the biggest impact we’ll see from climate change on energy systems,” he added.

The study found that climate change could increase energy failures by as much as 500% by 2050 if future climate conditions are not considered when siting and designing energy infrastructure.

Where wind and solar are built matters

The researchers found that locations best suited for renewable projects under future climate conditions can differ meaningfully from those identified using historic weather data.

“We are telling people where you put your wind and solar matters a lot for your ability to deliver energy when you need it,” Qiu explained.

“We need to think more about the when and where of adding renewables rather than only focusing on adding overall capacity,” he added.

In New England, climate-related disruptions increased the need for solar capacity and transmission lines close to major demand centers.

In Texas, risks were primarily driven by transmission constraints. Climate-informed designs prioritized wind farms in West Texas to better align renewable generation with future demand patterns.

Researchers seek faster planning tools

The current framework relies on expensive, high-resolution models, making it impractical for grid operators to use routinely today.

The researchers hope to develop faster models that could make climate-informed planning easier to apply.

“This study shows the opportunity and the need,” Howland says. “There are risks to not adapting our system, but if we do adapt our system, there could be big opportunities that are not costly.”

The study was published in Natural Energy.

The United States is facing nine major supply chain challenges that could delay the rehabilitation of its hydropower infrastructure, according to a new report from the National Laboratory of the Rockies (NLR).

The report has examined the difficulties hydropower operators encounter when upgrading or replacing large power transformers (LPTs).

These custom-built devices raise the voltage of electricity generated by hydropower turbines, allowing it to travel efficiently across long-distance transmission networks.

Their importance extends beyond hydropower, as more than 90% of the electricity consumed in the United States passes through high-voltage transformers at some stage.

Transformer deliveries can take five years

One of the most serious challenges identified in the report is the time required to procure new transformers.

A typical LPT rated at 100 megavolt-amperes takes approximately 2.5 to three years to acquire. Lead times for extra-high-voltage units can extend to five years because only a limited number of facilities worldwide can manufacture them.

Transformer prices sharply increased by approximately 80% in 2025 compared with 2021 levels, according to the report. A single unit can now cost as much as $10 million.

These pressures come as demand rises across the power sector. Although domestic manufacturing and assembly capacity has expanded in recent years, the report found that it has not grown quickly enough to meet current requirements.

US manufacturers also depend heavily on international suppliers. Approximately 80% of the raw materials used in domestic LPT production, including grain-oriented electrical steel and refined copper, are imported.

Changing trade conditions could further increase the baseline costs of copper and steel, adding uncertainty to transformer procurement.

Remote dams create transportation problems

Delivering an LPT to a hydropower facility presents another layer of difficulty. Many dams are located in isolated regions surrounded by rugged terrain, making them considerably harder to reach than conventional power facilities.

Transformers require specialized vehicles and heavy-duty trailers capable of carrying oversized loads. However, the United States has a limited domestic fleet of such equipment. Transportation routes must also be planned carefully and require special permits.

Permitting and regulatory procedures can produce additional delays, particularly when an operator must remove an existing transformer before installing its replacement.

Federal contracting rules and domestic-content requirements could also create near-term bottlenecks while US production capacity remains limited.

Report could guide domestic investment

The NLR study builds on the US Department of Energy’s broader Hydropower Supply Chain Gap Analysis, which previously identified five gaps affecting the domestic industry.

DOE’s Hydropower and Hydrokinetic Office funded the follow-on study to provide a closer examination of the LPT supply chain. Researchers used feedback from hydropower industry participants to define the nine challenges.

The findings could help shape federal policies, incentives, lending programs, workforce initiatives and investments intended to expand US transformer production.

Strengthening the domestic supply chain will be important for new construction as well as upgrades, refurbishments and relicensing work at existing facilities.

Hydropower currently provides more than 6% of total US electricity generation, making the availability of critical transformers essential to its continued contribution to the grid.

The report is titled as Large Power Transformer Supply Chain Gap Analysis and Domestic Content Strategies for Hydropower Rehabilitation.

Rune Technologies has unveiled TyrSense, a new line of ruggedized sensors designed to help military units track personnel, casualties, and supplies in contested environments where conventional communications may be unavailable.

The hardware marks the company’s first expansion beyond software, extending its TyrOS logistics platform directly to personnel operating at the tactical edge.

The initial TyrSense products use small, durable NFC tags that can be attached to personnel or cargo. When scanned during a handoff, the system records the location and status of the individual or pallet in TyrOS.

Unlike conventional tracking systems, the location is determined by the node where the scan takes place rather than through GPS.

The tags can also operate without an active cloud connection or satellite signal, allowing units to continue recording movements in denied, degraded, intermittent, or low-bandwidth environments.

Once connectivity returns, the collected information can synchronize with TyrOS.

Aim to replace manual military tracking

Military units continue to rely heavily on manual processes to account for personnel and supplies, according to Rune.

Information about available resources is often maintained through whiteboards, paper manifests, and manually compiled reports. These systems can become difficult to maintain when units relocate or lose network connectivity.

TyrSense is initially designed around three military workflows that commonly depend on manual reporting: personnel accountability, medical operations, and cargo tracking.

For personnel, scanning a tag allows units to update an individual’s duty status, reducing the need for the daily manual process of compiling accountability information.

In medical operations, the system can record each transfer of a casualty as the individual moves from the point of injury through successive levels of care.

For logistics, tags attached to pallets can register shipments as they move through different nodes across a military supply network.

Built to operate when battlefield networks fail

Rune has designed the sensors around conditions where military personnel may have limited or no communications access.

Because individual scans can be registered locally at the tactical edge, units can continue collecting information even during network disruptions.

“A tag has to work in a vehicle and with no signal, because that’s where these counts happen,” said Peter Goldsborough, Co-Founder and Chief Technology Officer of Rune Technologies.

“We built TyrSense so the scan registers at the edge and reaches TyrOS the moment a connection is there. The hardware is deliberately simple. The value is what TyrOS does with the data,” he added.

The approach allows TyrOS to receive updated information once connectivity becomes available rather than requiring personnel to maintain continuous access to remote infrastructure.

Rune plans sensors for fuel and water

TyrSense builds on the company’s deployment of TyrOS with operational military units over the past year.

During Balikatan 26 in the Philippines, the software provided tactical commanders with information on food, water, fuel, and personnel supporting thousands of troops across a Joint Task Force, according to the company.

Rune is also the sole logistics provider for the US Marine Corps’ Project Dynamis and is the only application selected by both prime contractor teams participating in the US Army’s Next Generation Command and Control program.

The company now plans to expand the TyrSense hardware portfolio beyond NFC-based tracking.

Creating a realistic 3D model from a simple image remains surprisingly difficult for artificial intelligence. 

While AI can generate impressive pictures, producing the precise CAD code engineers use to design aircraft parts, cars, and consumer products is far more challenging. Even a small mistake can make a design unusable. 

Now, researchers from MIT, IBM, and Red Hat have developed a framework that significantly improves the ability of AI systems to turn 2D images into functional CAD programs, producing more accurate designs while reducing inference computation by about 80 percent.

This new research work tackles one of the biggest obstacles in AI-driven engineering, which is the shortage of high-quality CAD training data. Existing image-to-CAD systems need vast numbers of examples linking 2D images to working CAD programs, but creating those datasets is expensive and labor-intensive. 

Instead of relying on more human-created data, the new approach teaches AI using the mistakes it already makes. “We want to obtain data augmentation that is informed by the model itself,” lead study author and AI researcher Giorgio Giannone said.

Turning failed designs into training material

The team’s framework, called Geometric Inference Feedback Tuning (GIFT), works with vision-language models that convert a 2D image and text description into Python code that can be executed in CAD software to create a 3D object.

To understand where a model struggles, GIFT asks it to solve the same CAD problem multiple times in parallel. Some outputs are correct, some are close to the target design, and others are much further away. 

According to the researchers, these ‘near-misses’ are the most valuable because they reveal exactly what the model does not yet understand.

The system executes each generated CAD program and compares the resulting 3D geometry with the target using a standard engineering metric known as Intersection over Union (IoU), which measures how closely two 3D shapes match. 

Rather than discarding imperfect outputs, GIFT identifies those that are close to correct and uses them to generate new training examples.

Giannone explained, “If we sample the model 10 times and it generates 10 correct answers to the same problem, then there is not much for it to learn. We care about the in-between cases, where the model might only solve the problem 50 percent of the time.”

Teaching the model how to fix its own errors

GIFT improves the model through two complementary techniques. The first, called GIFT-REJECT, keeps alternative CAD programs that generate the correct geometry even if the code differs from the original reference solution. 

Traditional AI training often assumes there is only one correct CAD program for each object, even though engineers can create the same part in many different ways. By exposing the model to multiple valid solutions, GIFT broadens its understanding of how CAD designs can be constructed.

The second method, GIFT-FAIL, focuses on imperfect outputs. The system takes a nearly correct 3D model generated by the AI, converts it back into an image, and pairs that image with the correct CAD program. By doing so, it creates new examples that effectively teach the model how to recover from specific mistakes.

More importantly, this process requires no human intervention. The framework automatically identifies errors, creates new training data, and uses that information to improve the model.

The researchers also designed GIFT around a technique known as inference-time scaling. Instead of retraining an entire AI model, which can be extremely expensive, the system improves performance during generation. 

Rather than discarding the extra computation spent searching for better CAD programs, GIFT converts successful and near-successful attempts into new training examples that can later be used to improve the model through fine-tuning.

Moreover, users can decide how much computation they want to spend depending on their budget and time constraints.

Better CAD designs with a fraction of the computing power

When evaluated on the GenCAD-Code dataset, models enhanced with GIFT achieved a 12 percent improvement in geometric accuracy compared with a strong supervised-learning baseline. 

The framework also reduced the inference computation needed to achieve comparable performance by about 80%.

“With GIFT, we started with geometry because with engineering problems, if the geometry of a 3D shape is not correct, nothing else will be correct, but there are many other aspects to consider,” Giannone said. 

The system’s ability to learn from its own mistakes could help accelerate product development by making AI-generated CAD models more reliable. 

Engineers could move more quickly from sketches to testable 3D designs, reducing development costs, shortening design cycles, and exploring a wider range of ideas without creating large amounts of new training data.

The researchers say this is only the beginning. So far, GIFT focuses on generating correct geometry, but future versions will aim to teach AI how to design parts that are not only accurate in shape but also easier to manufacture and better suited for real-world performance. 

The study is published on arXiv.

The US Air Force’s 142nd Wing has completed its first live bomb-dropping exercise with the F-15EX Eagle II, marking a key milestone in expanding the aircraft’s operational capabilities. In early May, pilots, maintainers, and munitions specialists traveled to Mountain Home Air Force Base in Idaho to carry out the training mission.

During the exercise, crews loaded 500-pound BDU-50 and 2,000-pound BDU-56 practice bombs onto the aircraft before releasing them over the Mountain Home bombing range. Although these training munitions closely resemble live bombs, they are filled with concrete instead of explosives, allowing crews to safely practice real-world weapons employment.

While the F-15EX has previously undergone weapons testing with the 85th Test and Evaluation Squadron at Eglin Air Force Base in Florida, this marks the first time an operational F-15EX unit has successfully loaded and dropped bombs as part of its mission readiness training.

F-15EX’s new combat role required months of pilot training

The F-15EX Eagle II is designed to do far more than its predecessor. Although it closely resembles the aging F-15C, the newer aircraft is built as a multirole fighter, capable of carrying out both air-to-air and air-to-ground missions. Unlocking that expanded capability, however, required months of preparation and specialized training across the 142nd Wing.

A key part of that effort was led by Maj. Jesse Loya, one of only three pilots in the wing with prior experience in air-to-ground operations. Together, the three completed a nine-month training program focused on the tactics and technical aspects of delivering precision-guided munitions. They then developed a training curriculum to teach the rest of the wing’s pilots the fundamentals of air-to-ground operations, laying the foundation for the unit’s first operational bomb-dropping mission.

According to Loya, the training at Mountain Home Air Force Base served as a stepping stone toward preparing pilots to safely employ air-to-ground weapons. The program focused on building the skills needed to handle ordnance while following a much shorter timeline than the traditional nine-month qualification course. 

He said the ability to complete the training so quickly reflected the experience and capabilities the pilots had developed throughout their careers, allowing the wing to accelerate the F-15EX’s transition into a multirole combat aircraft.

Weapons loaders play key role in prepping F-15EX

Preparing the F-15EX Eagle II for its first bomb-dropping mission required extensive training from weapons crews as well as pilots. Tech. Sgt. Tyler Phelps, one of the 142nd Wing’s experienced weapons loaders, traveled to Nellis Air Force Base in Nevada earlier this year to train with F-15E aircraft and earn certification to load bombs.

After returning to Oregon, Phelps and other certified Airmen trained the rest of the weapons team ahead of the Mountain Home mission. Seeing the bombs loaded onto the F-15EX marked the culmination of months of preparation and a major milestone for the 142nd Wing.

The training mission involved 10 sorties, during which crews dropped a total of 24 BDU-50 and 12 BDU-56 practice bombs. The operation was completed without any safety incidents, and all missions received satisfactory evaluations, marking a major step forward in the wing’s transition to the multirole capabilities of the F-15EX.

The next phase in the F-15EX’s air-to-ground certification process will involve loading and dropping live munitions. The crew is scheduled to conduct that mission at Nellis Air Force Base at the end of August, bringing the wing one step closer to full air-to-ground operational capability.

💾

British aerospace company Vertical Aerospace is set to conduct the first public electric vertical take-off and landing (eVTOL) demonstration flights at the upcoming Farnborough International Airshow. 

The event represents another milestone in the company’s journey toward certification and the eventual launch of commercial electric aircraft services.

The demonstrations come at a key stage in the company’s development, Vertical Aerospace noted. In recent months, the company has unveiled its Valo aircraft, completed successful piloted transition flights, strengthened its financial position, expanded its global technology partnerships, progressed its hybrid-electric defense strategy, and continued advancing its certification program.

Valo eVTOL moves closer to commercial operations

Following successful demonstrations of the aircraft’s core capabilities, Vertical Aerospace is now shifting its focus toward the next stage of development, with efforts centered on certification, industrialization and preparing for commercial operations.

Alongside public piloted flight demonstrations of its full-scale prototype, the company will also showcase its full-scale commercial Valo aircraft model, offering visitors a closer look at the technology behind its next-generation electric aircraft.

According to Stuart Simpson, CEO of Vertical Aerospace, the company arrives at Farnborough with the key building blocks for its next phase already in place. He said Vertical has transformed over the past year by launching its Valo aircraft, achieving piloted transition flights, strengthening its financial position and expanding its industrial ecosystem, while continuing to prove the technology.

“Every milestone we deliver reduces risk, strengthens our competitive position and brings us closer to building one of the world’s leading electric aerospace companies,” Simpson noted. A major highlight of Vertical Aerospace’s presence at the Farnborough International Airshow will be its planned flight demonstrations, which are set to mark the first public eVTOL aircraft demonstrations at the event.

Company targets key certification and production milestones after expanding flight tests

The company intends to conduct demonstration flights throughout the show, subject to approved flight conditions. The planned flights follow the UK Civil Aviation Authority’s recent expansion of Vertical’s Permit to Fly, allowing the company to carry out public demonstrations beyond its Flight Test Centre at Cotswold Airport for the first time.

Following the expanded flight approval, Vertical’s test pilots have successfully operated the aircraft across multiple locations, including RAF Brize Norton, RAF Benson, Blackbushe Airport and Farnborough Airport. These flights have expanded the company’s test program, providing valuable experience across different operating environments.

The company is also targeting several important program milestones that will support the development, production, and commercial rollout of its aircraft. These include completing the Critical Design Review (CDR) by the end of 2026, which will establish the certifiable design baseline for the aircraft and enable the production and testing of certification-conforming models. 

Furthermore, the UK firm plans to bring its early production aircraft assembly facility online in Q3 2026 and expand the Vertical Energy Centre in Q4 2026 as part of its industrialization efforts. In addition, Vertical expects to select a long-term turbogenerator supplier during 2026, ahead of beginning hybrid-electric flight testing in the first half of 2027.

💾