Skip to main content
frontier

Airhart Aeronautics begins installing production avionics hardware, ending prototype phase

By Elena Petrova

Why Airhart Chose In-House Avionics Manufacturing Over Outsourcing

General aviation cockpits still look like collections of boxes bolted to panels, with a radio here, a transponder there, and a glass display from one vendor talking to an autopilot from another through interfaces frozen in the 1990s. The result is a cockpit that demands a pilot manage the machine instead of the mission. A startup founded by a former SpaceX avionics engineer is betting that the only way out is to stop buying the boxes and start building the system.

Nikita Ermoshkin founded Airhart Aeronautics in 2022 after earning his pilot's license in 2020 and confronting the gap between modern avionics potential and general aviation reality. At SpaceX, he designed avionics for the most reliable rockets in the world. He saw no reason the same integration discipline couldn't apply to a four-seat airplane, where hardware and software are co-designed from the same spec, tested on the same bench, and updated over the same pipe.

The industry norm runs the other way. General aviation airframers typically source avionics from a handful of Tier 1 suppliers. Garmin, Avidyne, Honeywell, Thales, Cobham, GE Aerospace, and L3Harris dominate a market Fortune Business Insights valued at $108 billion globally in 2024, with North America holding roughly a third of that share. These suppliers deliver capable boxes, but the integration burden falls on the airframer or the installer. The pilot gets a panel stitched together from disparate product lines, each with its own menu logic, failure modes, and update cycle.

Airhart's rationale for vertical integration starts with that fragmentation. "Most airplanes today still use avionics designed around decades-old assumptions," the company wrote in a Y Combinator posting. "We think that's one of the biggest reasons general aviation remains difficult to learn and intimidating to use." The alternative is designing the flight displays, the autopilot servos, the engine management, the power electronics, and the harnesses as one system. This lets Airhart eliminate the interface debt that plagues legacy panels.

That philosophy extends to the manufacturing floor. Job postings for Airhart's production team read more like a rocket company than a traditional avionics shop: develop manufacturing processes for avionics computers, displays, power electronics, harnesses, and electromechanical assemblies; select and qualify vendors for PCB assembly, machining, sheet metal, cable assemblies, and coatings; own design-for-manufacturability reviews with electrical and mechanical engineers before hardware is released; design fixtures, tooling, jigs, and automated test setups. The goal is not just to assemble boards but to control the feedback loop between what the software needs and what the hardware can deliver.

The contrast sharpens when you look at the update path. Legacy avionics suites typically require physical shop visits for software changes, involving a mechanic, a laptop, and a certification paper trail. Airhart's architecture was built from the beginning for over-the-air updates delivered via Starlink connectivity, the same way a Tesla receives new driving logic overnight. That capability only exists because the flight software, the state-estimation engine, the display stack, and the communication hardware share a single development pipeline. No Tier 1 supplier contract includes that level of co-design access.

President Nate Thuli framed the safety stakes directly: "The technology and the training and the pathways to making commercial airline travel as safe as it is do not trickle into small aviation as well. Safety is a big detriment." He added that Airhart implements "a lot of logically based decisions into how this airplane will operate, automations and other ways of basically reducing the total chaos on that system, to really reduce the burden of this plane operating within that landscape."

Edward Story, co-founder of the Santa Monica-based Eco-Aviation Foundation International, described the broader shift as "the electronization of the cockpit." This is a trend Airhart is riding but also accelerating by refusing the standard supply chain. Cirrus Aircraft, Piper Aircraft, and Garmin Avionics have each added assisted-safety features: whole-airframe parachutes, autonomous emergency landing, envelope protection. Honeywell International Inc. is pursuing AI for prediction and route optimization. But each of those features lives inside a supplier's roadmap, not the airframer's. Airhart's bet is that the only way to make the automation feel like a co-pilot instead of a checklist is to own the stack.

The company raised $4.5 million in venture capital in October 2024 and opened its first design and innovation center in Long Beach. This location is a deliberate echo of the "Space Beach" cluster of launch and satellite companies, housing the engineering and manufacturing teams side by side. The prototype has been flying since its public reveal at Santa Monica Municipal Airport in 2024, developed in collaboration with South Africa-based Sling Aircraft. The next step, detailed in the following section, is the installation of production avionics hardware and the core software architecture that makes the integrated platform manufacturable at scale.

From Prototype to Production: The Milestone That Signals Scale Readiness

Airhart Aeronautics announced in late January that it had deployed the core software architecture underpinning its first production avionics panel. The company explicitly frames this as the transition from development to deployment. The announcement, carried by Airframer, marked the first time a general aviation startup had fielded an integrated, real-time state-estimation engine as the shared foundation for an entire avionics stack. Aviation Week confirmed weeks later that Airhart had moved beyond software deployment and begun installing production hardware in aircraft.

The hardware itself carries the weight of that claim. In a walkthrough recorded in August, the company showed the Aircore unit: two 14.1-inch high-definition touch displays housed in a single 31-inch panoramic assembly, backed by two fully redundant flight computers. Each computer can run the complete Aircore platform independently, delivering fault tolerance at the compute layer. Behind the displays, Airhart has eliminated the traditional wiring harness entirely. Every component, down to the autopilot servos, is developed in-house, a level of vertical integration that lets the team trace fault tolerance to the ball bearings inside each servo.

"This is not about new math," said Nikita Ermoshkin, co-founder and chief technology officer. "The breakthrough is architectural. We made real-time state estimation the foundation of the entire avionics system, not a hidden module feeding a display, but the shared brain of the airplane."

That engine ingests high-rate data from inertial sensors, GPS, pitot-static systems, magnetometers, and other inputs, fusing them into a single, resilient picture of attitude, position, velocity, acceleration, and three-dimensional orientation. Sensor errors, drift, and failures are detected and corrected inside the estimation framework itself. Airframer noted this design choice "dramatically improving robustness compared to legacy systems." The result is a shared source of truth that the automation stack, the displays, and the pilot interface all reference without translation layers.

Nate Thuli, Airhart's president, called the milestone "a defining moment" and emphasized that the company is not merely delivering new hardware but "introducing a fundamentally new way for general aviation aircraft to understand themselves in flight." The flight test campaign that followed validated the integrated platform in active flight, moving the system out of the lab and into the airframe it was designed for.

Early installation data underscores the manufacturing payoff. One builder who replaced a legacy Garmin-based suite with the Aircore panel reported a weight reduction of roughly 20 pounds. This is a direct consequence of consolidating discrete boxes, harnesses, and interconnects into a single integrated assembly. The simplification extends to the build line: fewer parts, fewer connectors, and a harness-free architecture reduce assembly labor and inspection points.

Pilot feedback from the validation flights reinforces the production readiness. A test pilot accustomed to conventional glass cockpits described the initial impression as "nothing like what I'm used to," then added that within feet of liftoff "it felt at home and natural." That transition, from unfamiliar to intuitive without a training plateau, is the practical signal that the hardware and software have matured together.

The milestone also unlocks the over-the-air update path that legacy GA avionics cannot support. Because the Aircore platform connects via Starlink, charts, databases, and software revisions push automatically, with no USB sticks, no shop visits, and no configuration drift across a fleet. For a company preparing to scale beyond experimental and light-sport categories into Part 23 certification under the FAA's MOSAIC rulemaking, that update mechanism is as much a production tool as a customer feature: it lets Airhart iterate on the production line without freezing a hardware baseline.

Four years from founding to first production install is compressed by aerospace standards. The architecture was built to support that pace, being modular, redundant, and software-defined from the servo up. The installation of production hardware is the evidence that the model holds.

How Integrated Avionics Enable Over-the-Air Updates and AI-Assisted Flight

Airhart Aeronautics built its avionics platform around a single architectural decision: treat every component as part of one networked system rather than a collection of boxes bolted to a panel. This includes displays, flight computers, radios, and servos. That integration is what makes over-the-air updates practical, and it is what lets artificial intelligence move from novelty to cockpit utility.

The company's Aircore platform centers on dual 14.1-inch HD displays paired with two fully redundant flight computers, each capable of running the entire avionics stack on its own. Behind the panel, software-defined radios and custom-developed autopilot servos connect through a unified network architecture that Airhart developed in-house, down to the servo level. President Nate Thuli described the approach as a ground-up redesign of a vertically integrated avionics platform, one that eliminates the wiring harness and complex CAN bus daisy-chaining that legacy general aviation systems depend on.

That architectural shift matters because legacy GA avionics treat software as fixed at the factory. Garmin-based systems, which Airhart has removed from customer aircraft, require physical access to update charts, databases, or firmware. Pilots plug in USB drives, download files, and manually load them through layered menu systems. Airhart's system treats software as continuously evolvable. Updates arrive free over Starlink for launch customers, included for at least two years, and the company is working on pricing beyond that initial period. The system automatically pulls the latest features, charts, and databases without pilot intervention.

The continuous update model feeds directly into AI-assisted flight. Airhart uses an onboard language model to transcribe radio communications in real time, even through garbled or hard-to-hear audio. When air traffic control calls a pilot's tail number, the system highlights that exchange in bold on the display. Readback instructions, such as headings, altitudes, and frequencies, surface as context-aware callouts that the pilot can confirm and load directly into the autopilot with a single interaction. Thuli compared the transition to the shift from manual transmissions to automatic in automobiles, where technology made previously essential skills largely unnecessary for safety.

The AI layer extends beyond radio transcription. The system tracks pilot behavior across flight phases, collecting data on proficiency, checklist usage, communication patterns, and maneuver execution. That builds a picture of pilot readiness that adapts to each individual. In an emergency, the platform presents the relevant checklist automatically rather than burying it in menus, and can even offer to execute the procedure itself. Thuli said the system can prevent the kind of simple mistakes that accident analyses consistently trace to loss of control.

Context awareness drives the automation. The platform monitors phase of flight, aircraft state, pilot inputs, and environmental conditions to surface only the information the pilot needs at that moment. A rubber-band methodology for turn coordination and vertical movement keeps the pilot contextually coordinated with the aircraft's state. The system can prompt a pilot to practice a specific approach in a simulator before a flight, then execute the mission efficiently once airborne.

This is not a retrofit bolted onto existing hardware. Every component on the Airhart platform, from the autopilot servo to the display unit, uses the same integrated architecture. That vertical integration, combined with the Starlink-connected update pipeline, lets features ship as software layers added after launch rather than requiring new hardware installations. Thuli said the company is already working on additional features that will arrive as post-launch software updates, building on the context-aware foundation established during the four years of development that preceded the production hardware installation.

The Impact on Pilot Accessibility, Safety, and Training Burden

Airhart's avionics suite does not just replace legacy instrumentation. It rewrites what a pilot has to manage in real time. The context-aware system monitors phase of flight, aircraft state, pilot inputs, and environmental conditions to surface only the information relevant to the current moment. Where traditional GA panels force pilots to scan dozens of gauges and navigate nested menus while simultaneously flying the airplane, Airhart collapses that task list into a single, prioritized display.

That reduction in task saturation shows up most clearly in the cockpit workload the system removes. One-button engine start, automated run-up assistance, and engine management that brings the entire powerplant online with a single press eliminate sequences that have historically required pilots to reference checklists, manipulate multiple switches, and cross-check instruments under time pressure. Emergency checklists, which legacy systems bury inside layered menu trees, appear automatically when the aircraft senses an anomaly, and the system can execute them on command.

The mental bandwidth freed by those automations becomes the foundation for accessibility. In the company's simulator, test pilots can toggle the simplifying features on and off, and the difference is tangible. One evaluator described the moment the assistance engaged as a literal "lightness," where the burden of micro-tasks lifted enough to let the pilot focus on flying rather than managing systems. That same evaluator, a self-described die-hard who had spent an entire career with traditional Garmin-style displays, reported that within ten feet of simulated rollout he understood everything the new interface presented. This admission underscores how aggressively Airhart has borrowed design language from automotive UIs to make the transition intuitive for non-pilots.

Radio work, long one of the most cognitively demanding elements of flight, gets handled by an onboard language model that transcribes ATC communications in real time. When a controller calls a tail number, the transcription highlights the relevant exchange in bold, giving the pilot a visual cross-check against the audio. Clearance items like "270, 3,500" auto-populate the autopilot, cutting down the manual programming that eats up attention during busy departure or arrival phases. The system can do this either through Starlink, which Airhart includes for at least two years on launch customer aircraft, or through local processing that keeps it functional even without connectivity.

Training burden drops alongside workload. The adaptive pilot development platform generates personalized training insights from real-world flying behavior, tracking procedure usage, checklist adherence, communication patterns, and proficiency trends over time. That data feeds into an assisted pilot readiness system that compares individual pilot proficiency against aircraft capability and planned flight conditions, producing a dynamic preparedness picture before takeoff. Instead of a static currency check, pilots get continuous feedback that tells them not just whether they are current, but whether the specific mission they are about to fly sits within their demonstrated envelope.

All of this is grounded in internal flight testing rather than theoretical design. The modified South African Sling TSi that Airhart gutted and rebuilt with its full avionics stack served as the testbed for these features, including the removal of roughly 20 pounds of hardware compared to the Garmin system it replaced. That weight savings matters not just for performance but for installation complexity, a barrier that has historically priced many experimental and light sport aircraft owners out of modern avionics upgrades.

The company's three-phase roadmap makes clear that accessibility is not a side effect but the central metric. Phase one delivers the context-aware avionics suite, phase two introduces fly-by-wire flight controls that sever the physical link between control stick and flight surface, and phase three couples both into a clean-sheet aircraft design built around simplified operation. Each phase layers on the last, but each also stands on its own as a workload reduction tool.

Airhart's approach directly challenges the assumption that GA safety improvements must come at the cost of increased pilot skill requirements. By pushing complexity into the avionics architecture rather than onto the pilot, the company is betting that the barriers to entry, including cost, training time, and cognitive demand, can be reduced without compromising the fundamental competency needed to fly. Early feedback from test pilots suggests that bet is paying off, at least in the simulator and in the modified Sling that has already flown.

What This Means for the Future of Avionics Supply Chains in GA and Beyond

Airhart's in-house model is already drawing attention from outside the experimental market. On June 1, 2026, the company announced a collaboration with Polish startup Draco Aircraft to explore integrating its avionics and flight control system into Draco's future general aviation platforms, according to Aviation Week. That partnership signals that Airhart's vertically integrated approach, designing everything from autopilot servos to displays internally, is being tested as a blueprint for other manufacturers that want to move faster than traditional Tier 1 suppliers allow.

The contrast with legacy supply chains is stark. General aviation avionics today still largely follows a model where airframers license or purchase certified systems from established players like Garmin, Honeywell, or L3Harris. Those companies optimize for certification compliance and broad compatibility across hundreds of airframes, which makes their update cycles slow and their interfaces conservative. As McKinsey noted in a 2025 report on aerospace manufacturing, most defense and aerospace contractors are struggling to match the pace of technological change, with only a select few breaking old paradigms to achieve rapid rate increases.

Airhart's architecture sidesteps that bottleneck by treating avionics as a single, unified system rather than a collection of boxed components. Its dual 14-inch touchscreen displays draw from a real-time state-estimation engine that ingests data from inertial sensors, GPS, pitot-static systems, and magnetometers simultaneously. That integration lets the system detect and correct sensor errors within the estimation framework itself, something legacy systems cannot do without external intervention. Nate Thuli, Airhart's president, called it a "defining moment" when production hardware began shipping, saying the architecture enables a new level of situational awareness and pilot protection previously unseen in general aviation.

For smaller GA manufacturers or experimental aircraft builders, that approach offers a path to differentiation without reinventing the wheel. Instead of sourcing discrete components and hoping they communicate reliably, they can adopt a platform designed for continuous improvement through over-the-air updates, including AI-powered features like live radio transcription and automated ATC readbacks. As one longtime observer put it during Airhart's Oshkosh debut in July 2026: "For the first time I think some of the major avionics players have the next generation of competition that we've been needing forever."

But scaling that model beyond niche or low-volume production presents real limits. High-rate manufacturing demands supply chains optimized for volume, not iteration speed. Airhart's current focus remains squarely on experimental and light sport aircraft, where regulatory flexibility allows meaningful innovation at a pace the certified world cannot match, as Flying Finance reported in June 2026. The company's three-phase roadmap, covering avionics first, then fly-by-wire, then a clean-sheet aircraft entering development in 2027, reflects a deliberate expansion into higher-complexity domains while staying within manageable production volumes.

Defense contractors and larger aerospace firms may borrow concepts from Airhart's playbook, particularly its emphasis on software-defined systems and human-centered design, but replicating full vertical integration at scale would require massive capital investment and cultural shifts. The real influence lies not in copying Airhart wholesale, but in adopting its core principle: treat avionics as an evolving service, not a static product. That mindset alone could pressure legacy suppliers to abandon incremental upgrades in favor of architectures built for continuous delivery, a change long overdue in an industry where cockpit technology has changed little in decades.


Working in frontier tech? Zero G Talent tracks the openings: see every open ASML role, browse frontier tech jobs, openings at Stripe, and the people building the field.

Ready to Start Your Space Career?

Browse frontier jobs and find your next opportunity.

View frontier Jobs