THE APEX TIMES
Lockheed Martin and the U.S. Navy test an AI “sonar expert” designed to retrain in minutes during RIMPAC 2026
The companies demonstrated SensorMAX, a rapidly retrainable AI/ML system that analyzes subsurface acoustic data, supports faster target detection, and pushes small model updates back to operators over the air.
Lockheed Martin said it demonstrated an AI-powered antisubmarine warfare capability with the U.S. Navy during RIMPAC 2026, using two MH-60R helicopters off the coast of Hawaii to conduct missions against an undersea threat. The exercise centered on SensorMAX, described by the contractor as a “rapidly retrainable” AI/ML sonar expert designed to help aircrew detect and track underwater contacts sooner than traditional approaches.
In Lockheed Martin’s description of the test, SensorMAX monitors spectral energy, including subsurface acoustic data, to assist crews with rapid threat detection. The company said the system is built to learn from new information during a mission, aiming to improve situational awareness as conditions and targets change.
The system’s workflow, as outlined by Lockheed Martin, is built around a closed loop of capturing audio-relevant data, labeling a target noise signature when operators identify a new contact, retraining the model, and redeploying an update to the aircrew. SensorMAX ingests subsurface acoustic data from a network of sensors and then provides a continuous data stream to aircrew and a channel to the ground station.
A core feature of the demonstration is its ability to update the AI model through an over-the-air (OTA) capability. Lockheed Martin said that when operators label an identified noise signature, the system retrains the model on the new data and sends a small update packet back to the aircrew. The company added that retraining takes under five minutes per iteration, positioning the technology for near-real-time improvements rather than longer development cycles.
During RIMPAC 26, Lockheed Martin said the demonstration used two MH-60R helicopters equipped with its AI/ML system to conduct a series of antisubmarine warfare missions against an undersea threat. While the release does not specify target types, engagement outcomes, or performance metrics, it frames the demonstration as a proof that integrated platform use, combined with AI-enabled acoustic classification, can speed the process of finding and tracking underwater threats.
Lockheed Martin’s update also emphasized that SensorMAX is intended to be platform-agnostic, meaning it is designed so that the “intelligence and classification” can be leveraged with different acoustic sensor inputs. The company said this approach is aimed at expanding a protective ring around high-value assets, and at accelerating what it calls the ASW kill chain, the sequence of actions from detecting a contact to classifying it and enabling follow-on decisions.
In comments included with the release, Devon Rodgers, vice president and general manager of Lockheed Martin’s Undersea Mission Systems, said the demonstration showed how the company is working to accelerate capability delivery and integrate commercial technology into mission-focused defense systems. Rodgers also said Lockheed Martin is investing its own resources to mature AI capabilities and deliver technology the Navy needs for “tomorrow’s battles.”
Defense industry observers have long noted the difficulty of antisubmarine warfare, where underwater sound propagation, background noise, and varying ocean conditions can make it hard for classifiers to remain accurate over time. Lockheed Martin’s closed-loop approach is designed to address that challenge by updating the model as new target signatures are identified, effectively turning operator feedback into faster model improvement during operations.
Still, important details were not disclosed in the company’s announcement. Lockheed Martin did not provide quantitative results such as detection probability, false-alarm rates, latency from sensor data to operator display, or how performance compared against traditional non-AI methods during the RIMPAC missions. The release also does not describe integration timelines for fleet use or the specific interfaces between SensorMAX and the helicopters’ existing antisubmarine systems.
Why It Matters
- If the retraining and OTA update approach performs as intended, it could reduce the time gap between newly observed undersea signatures and improved classification accuracy for operators.
- The demo highlights a shift toward AI-enabled sensing and decision support in undersea warfare, where fast learning can be critical in changing acoustic environments.
- Lockheed Martin positioned the technology as platform-agnostic, which could make it easier to scale the capability across different helicopters and sensor suites.
- The emphasis on a faster “kill chain” underscores the Navy’s interest in accelerating the end-to-end process from detection to actionable targeting in antisubmarine warfare.
Sources
Key Facts
- Lockheed Martin and the U.S. Navy demonstrated SensorMAX, an AI/ML system for antisubmarine warfare, during RIMPAC 2026 off Hawaii.
- SensorMAX is designed to monitor spectral energy, including subsurface acoustic data, to assist crews with rapid threat detection and tracking.
- The system uses a closed-loop workflow: capture data, label a target noise signature, retrain the model, and redeploy updates to the aircrew.
- Lockheed Martin said model updates can be pushed over the air, with retraining taking under five minutes per iteration.
- Two MH-60R helicopters equipped with Lockheed Martin’s AI/ML system conducted antisubmarine missions as part of the exercise.
Defense Related
Telecom comparison turns on profitability pace versus leverage: AT&T’s margin jump, Verizon’s debt load
A recent market comparison highlights how AT&T and Verizon can reach investor appeal through different routes, with AT&T showing a sharp boost in net margin while Verizon carries heavier balance-sheet leverage, even as both distribute dividends.
Anthropic agrees to a $35 billion cloud computing deal tied to Nvidia-backed Lambda, report says
Anthropic PBC is reportedly moving to lock in large-scale compute capacity through a major multi-year arrangement with Lambda, a cloud provider backed by Nvidia. Terms and timelines were not fully disclosed in the report.
AMD has tended to fall in September, but market history is only part of the story
A review of the past decade points to a recurring pattern for AMD in September. The stock has declined in eight of the last 10 Septembers, though broader market seasonality appears to explain only some of the weakness.
Apple escalates claims against OpenAI, alleging evidence destruction in trade-secrets fight
In a new court filing, Apple accused OpenAI of actively destroying evidence tied to a trade-secrets dispute involving a former iPhone engineer. The company also pressed claims tied to alleged downloads of confidential information.
Duolingo shares jump after results point to steady user momentum, according to Yahoo Finance
A Yahoo Finance report highlighted that Duolingo’s second-quarter revenue rose 18% year over year, using the framing of a “Netflix-like comeback” after a period of volatility in the online learning category.
Netflix confirms production of Korean series “Materesa (WT),” led by “Queen of Tears” director and writers behind “The East Palace”
The streamer says its next Korean mystery drama, centered on a cold-blooded criminal psychologist who probes unsolved murders, is in production and has set a cast for “Materesa (WT).”
FTC and 22 States Sue Amazon, Alleging It Secretly Marked Up Ads Shown to Marketplace Sellers
The federal competition regulator and a coalition of states claim Amazon undercut third-party sellers on its platform by allegedly embedding surcharges into advertising terms.
Deere shares gained as market focused on a jump in profits
Investors appeared to bid up Deere & Company after a market report pointed to sharply higher profit expectations, underscoring how quickly sentiment can turn in farm equipment when earnings outlooks move.
Baird lifts Deere to Outperform, citing potential agricultural recovery and raises target to $800
The firm upgraded Deere & Company to Outperform from Neutral and increased its price target to $800 from $640, pointing to improving conditions in agriculture as a key catalyst.
Venezuela’s energy reopening talks could create upside for Chevron and GE Vernova, but agreements still face major hurdles
Companies including Chevron and GE Vernova are reportedly among bidders or potential partners that could benefit if final deals for Venezuela energy projects move forward. Still, the process appears unfinished, and key risks around sanctions, contracts, and execution remain.
