AI news ·
Archer teams with NVIDIA to bring IGX Thor AI to Midnight eVTOLs, Hawthorne debut ahead of the 2028 LA Olympics
Archer is integrating NVIDIA's IGX Thor to assist pilots with real-time sensing and path evaluation. First trials roll out at HHR ahead of a commercial push.
Archer partners with NVIDIA on integrated AI-assisted flight operations tech
Archer Aviation has selected NVIDIA's IGX Thor platform to support future flight operations. The AI-driven system will assist pilots and operators with real-time situational assessment and decision support in safety-critical conditions.
IGX Thor is NVIDIA's most powerful safety-capable AI computing module. Archer plans to integrate it into future iterations of the Midnight eVTOL and its proprietary flight management systems to sense the aircraft's environment, process flight-path data in real time, and reduce pilot workload in busy airspace.
This stack also lays the groundwork for semi-autonomous and autonomous operations over time.
The first real deployment will be at Hawthorne Municipal Airport (HHR) in the Greater Los Angeles area. Archer acquired the airport in 2025 to test aircraft in live conditions ahead of a planned commercial rollout before the 2028 LA Olympics. The company announced the NVIDIA partnership during CES in Las Vegas (January 6-9, 2025).
Why this matters for operations
- Workload and throughput: Onboard perception and real-time path evaluation can shorten decision cycles for pilots and dispatchers, supporting higher sortie rates without sacrificing safety margins. See further context on AI for Operations.
- Standardization: Consistent AI-aided advisories can reduce variability across crews and shifts, tightening SOP adherence and improving repeatability.
- Scalability: As vertiport networks grow, decision support will help manage dense schedules, tighter turn times, and dynamic reroutes.
Immediate to-dos for ops leaders
- Update SOPs and CRM: Define when crews accept, challenge, or override AI advisories. Clarify roles between pilots, dispatch, and maintenance control for AI-flagged events.
- Safety case and failure modes: Run FMEA on the AI compute path, including degraded modes, fail-safe behavior, and independence from primary flight controls.
- Data governance: Establish retention and access rules for sensor feeds, AI inferences, and event logs. Treat it like FOQA for eVTOL, with clear de-identification and audit trails.
- Maintenance readiness: Prepare spares, test equipment, and technicians for high-compute avionics. Track MTBUR for the compute module and thermal management checks at turnarounds.
- Cybersecurity and updates: Lock down secure boot, SBOM tracking, and patch windows. Align OTA update playbooks with airworthiness approvals and operational availability targets.
- Training and simulation: Build scenario-based training where AI advisories help, conflict, or degrade. Include dispatch and MOC in the loop, not just flight crews, and apply Prompt Engineering practices to design advisory interfaces and prompts.
Key questions to pressure-test with your vendors and team
- Which functions are advisory versus commanding, and what are the handover rules?
- What are the latency budgets and compute margins under worst-case sensor loads?
- How is the safety case structured (DAL allocation, independence, verification evidence)?
- What's recorded, for how long, and how quickly can we retrieve it after an event?
- What is the plan for software updates, rollback, and certification of changes without grounding fleets?
- How were human factors validated for alerting, explainability, and crew workload?
KPIs to track from day one
- Dispatch reliability and delay minutes attributable to AI systems
- Advisory acceptance rate and false/late alert rate
- Time to detect and resolve hazards flagged by AI
- Software update success rate and time-to-return-to-service
- Mean Time Between Unscheduled Removals (compute module and associated sensors)
- Energy and thermal headroom during peak operations
What to watch next
- HHR operational trials: Procedures for vertiport turnaround, ATC coordination, and real-time reroute logic will set the playbook for other cities.
- Certification path: Evidence for safety-capable AI, change management for software, and clear delineation between advisory and control functions.
- Scaling constraints: Supply chain for high-performance compute, thermal management in hot weather ops, and maintenance throughput.
- Autonomy increments: Where semi-autonomous features enter service first (e.g., taxi, takeoff/landing aids, contingency management) and how crews supervise them.
If you're building team capability around AI oversight and automation, you may find this resource useful: AI courses by job role.