The Deal
On July 29, 2026, TPIsoftware, the Taiwan-based enterprise software provider founded in 2005, signed a global reseller agreement with Juxta, a San Francisco startup from Y Combinator's summer 2025 batch, folding Juxta's Universal Positioning System, a software-only layer that turns inertial measurement units into sub-meter positioning engines, into TPIsoftware's integrated AI platform for deployment across Vietnam, Singapore, Thailand, and Japan. The partnership targets a direct consequence of GPS architecture: satellites cannot see inside tunnels, under dense canopy, or through deliberate jamming. For decades the workaround was more hardware: beacons, repeaters, ground stations. Juxta asked a different question: what if the device itself could figure out where it is, using only the sensors already inside it?
"Partnering with TPIsoftware strengthens our regional presence, allowing us to tap into their integration capabilities and established networks to enter broader markets across APAC," said John Ferrara, Juxta's founder and chief executive. Ferrara's path to the problem was personal; both parents served as U.S. Navy officers, and their experiences with GPS limitations in critical moments shaped the company's founding thesis in January 2024. Juxta began in healthcare logistics, tracking patients and staff through hospital corridors where satellite signals don't reach. By January 2025 the team had built an IMU simulator that let them train positioning models without collecting months of real-world data. That simulator became the foundation for UPS, which launched publicly in September 2025. NODE1 asset tags followed in January 2026. First commercial deployments began in March.
TPIsoftware's general manager, Yilan Yeh, framed the deal as a compliance play as much as a capability play. "The strategic collaboration extends our reach to industries with higher compliance standards while enabling deeper integration of our proprietary AI solutions with Juxta's positioning capability," Yeh said. "By using Juxta's accurate positioning and anti-jamming techniques, we can further support customers in healthcare, disaster prevention and logistics sectors in Taiwan and those beyond APAC with on-premises AI deployment for enhanced data sovereignty, security and resilience."
The combined pitch is specific: UPS delivers one-meter accuracy indoors, outdoors, underground, and in GPS-denied environments without beacons, anchors, cameras, or new infrastructure. Deployment takes under an hour, fully remote. The system continues tracking offline and syncs when connectivity returns. Juxta claims up to five times lower cost than traditional tracking installations. Those claims have been tested in government, defense, logistics, and healthcare settings, including continuous positioning inside tunnels and navigation for unmanned systems in mountainous terrain where signal reception is limited or disrupted.
What the announcement doesn't spell out is how the reseller economics split, whether TPIsoftware takes equity, or which specific defense programs have already evaluated the combined stack. The PR Newswire release describes "field-tested and deployed across government, defense, logistics, and healthcare environments" but names no defense customers. That silence is notable for a partnership explicitly targeting security sectors.
How It Works: Ground‑Based Positioning Without GPS
The core of Juxta's Universal Positioning System rests on a sensor-fusion architecture that sidesteps satellite signals entirely. UPS is built around Inertial Measurement Unit sensors paired with advanced machine learning to deliver real-time navigation, asset tracking, and workflow automation, without the receivers, beacons, or cameras that conventional indoor positioning systems require. The IMU provides raw inertial data: accelerometers and gyroscopes measure linear acceleration and angular velocity, which onboard algorithms integrate to estimate position, orientation, and velocity through dead reckoning. Left alone, that integration drifts. The machine-learning layer corrects it.
Juxta describes the result as a "hardware-free, offline-capable platform that transforms any environment into an intelligent, continuously trackable space." The phrase "hardware-free" refers to the absence of fixed infrastructure: no ultra-wideband anchors, no Bluetooth beacons, no visual fiducials mounted on walls. The only hardware is the sensor package carried by the person, asset, or robot being tracked. That package runs its positioning computation locally, which makes the system offline-capable and inherently resistant to jamming or spoofing: there is no external signal to disrupt. The PR Newswire release notes that UPS "supports sub-meter tracking accuracy anywhere on Earth even in circumstances where GPS is denied and the signal is obstructed."
The anti-jamming claim follows directly from the architecture. GPS vulnerabilities, including limited coverage, signal jamming, and spoofing, stem from reliance on weak satellite transmissions in the L-band. UPS does not listen for those transmissions. Its IMU sensors are self-contained; the ML models that correct drift are trained on motion patterns and environmental signatures captured during operation, not on external radio references. That makes the positioning solution passive from an RF perspective: it emits nothing an adversary can detect or interfere with, and it receives nothing an adversary can deny.
TPIsoftware's contribution is the integration layer. The Taiwanese firm, founded in 2005, specializes in proprietary enterprise solutions for mission-critical systems, artificial intelligence, cybersecurity, and sustainability management. In this partnership, TPIsoftware provides on-premises AI deployment that wraps Juxta's positioning output into the data-sovereignty, security, and resilience requirements of defense and critical-infrastructure customers.
What the research does not specify is the exact ML architecture, whether it uses recurrent networks, transformer-based sequence models, or classical probabilistic filters augmented with learned corrections, nor the IMU grade (tactical, navigation, or consumer). It also does not quantify drift rates over time or distance without external correction, or how the system handles long-duration static periods where IMU bias stability becomes the limiting factor. Those details remain proprietary. What is documented is the system-level outcome: infrastructure-free, sub-meter positioning that operates where GPS cannot, with a software stack designed for on-premises deployment in regulated sectors.
Why It Matters for Defense Now
The partnership lands at a moment when GPS vulnerability has moved from theoretical concern to operational crisis. Modern conflicts, such as Russia–Ukraine and Israel–Hamas, have demonstrated that satellite signals can be jammed, spoofed, or simply lost in contested environments, severely impairing military effectiveness according to IDSA analysis from October 2024. For defense installations and secure facilities, the stakes are immediate: navigation and tracking must function when the sky goes dark.
Juxta's Universal Positioning System has already been field-tested and deployed across government, defense, logistics, and healthcare environments where reliable positioning is critical. The system delivers sub-meter tracking accuracy anywhere on Earth even when GPS is denied or signals are obstructed. That claim matters because it addresses the core gap: continuous positioning inside tunnels, indoor and underground spaces, and navigation for unmanned systems operating in mountainous or remote areas where signal reception is limited or disrupted.
For military bases, the immediate effect is resilience. A base's internal logistics, including supply movements, personnel tracking, and autonomous vehicle convoys, can maintain positioning continuity when satellite links are severed. Secure facilities gain the ability to monitor full-scale operations indoors and underground, zones where GPS never worked to begin with. Autonomous security patrols, a growing priority for perimeter defense and critical infrastructure protection, can navigate predefined routes in GPS-denied conditions without the infrastructure cost of beacon networks.
TPIsoftware's integration layer is what makes this deployable at scale. Both statements signal that the technology has moved past prototype into integration-ready form, a distinction that matters for defense procurement cycles.
The performance claims are specific: sub-meter accuracy, field-tested across multiple sectors, operational in tunnels and underground. What's absent from the public record are named pilot programs at specific bases or quantified results from live exercises. The announcement says "deployed across government, defense" without naming agencies or locations. That silence is typical for defense-adjacent deployments but leaves the magnitude of current adoption unmeasured.
Industry Reaction: Wait and See
Beyond the principals, the public record as of late July 2026 shows no on-the-record statements from U.S. defense primes specifically addressing the Juxta–TPIsoftware tie-up. Lockheed Martin's August 2026 news flow centered on its own NetSense™ airspace-protection demonstration with Verizon, Keysight, ODC and its Astris AI subsidiary; the Kodiak containerized launcher live-fire test; NGI Stage 2 motor static firing; MDA modeling-and-simulation award; Strigo modular missile-technology launch; and hypersonic payload-delivery and composites teaming agreements. None of those releases reference satellite-free terrestrial positioning or the Juxta–TPIsoftware arrangement. The same holds for allied-nation procurement offices: Australia's Submarine Agency awarded Lockheed a AUD 26.5 million combat-systems integration contract for the Virginia-class fleet in August 2026, but no allied defense ministry has publicly evaluated UPS for base security or autonomous patrol fleets.
Competing positioning firms have likewise been silent in open channels. Companies developing alternative GPS-denied navigation have not issued comparative assessments. The absence of public pushback or endorsement is typical for early-stage dual-use partnerships where evaluation cycles run 12–24 months before prime contractors commit integration engineering resources. The technical bar for adoption is interoperability with existing GPS-augmented architectures, such as M-Code receivers, SAASM modules, and the Army's Mounted Assured PNT System (MAPS), not raw positioning accuracy alone. Until Juxta and TPIsoftware publish interface control documents or demonstrate MAPS-compliant data formats, primes will treat UPS as a technology watch item rather than a procurement candidate.
The silence from allies is similarly structural. Taiwan's research institutes evaluate indigenous PNT solutions under national defense autonomy roadmaps, but their assessment reports are classified. Japan and South Korea pursue quantum-inertial and terrestrial beacon alternatives. None has publicly benchmarked UPS. NATO maintains a technology registry, but inclusion requires a formal submission and security accreditation process that typically follows, not precedes, a prime contractor's integration decision.
In short, the only documented reactions are the two CEOs' own framing. The defense-industrial base, allied procurement agencies, and competing PNT vendors are in a wait-and-see posture, monitoring technical disclosures, interface standards, and the first prime-contractor integration contract before committing public positions.
Ripple Effects for Autonomous Vehicles and Robotics
The partnership's most consequential ripple effect may land far from military bases. Juxta's Universal Positioning System has already been applied to warehouse and logistics operations, per the companies' own disclosure, and the technical architecture — IMU sensors fused with machine learning, no beacons, no cameras, no GPS receivers — maps directly onto the pain points that have stalled robotics deployments in GPS-denied environments for years.
Consider the warehouse robot problem. Autonomous mobile robots in fulfillment centers typically rely on lidar SLAM, floor markers, or ultra-wideband beacons. Each approach carries infrastructure cost: beacons need power and maintenance; lidar struggles with dynamic obstacles and featureless corridors; floor markers degrade. Juxta's UPS claims sub-meter accuracy without any of that infrastructure. If the field-tested performance holds (the PR notes deployments across logistics environments), a robot fleet could localize itself purely from inertial data and learned motion models, cutting capex and eliminating a class of failure modes tied to beacon outages or map drift.
The same logic extends to self-driving vehicles operating in tunnels, parking structures, and urban canyons. GPS dropout in these scenarios isn't an edge case; it's a daily operational reality. Current autonomy stacks handle it with high-definition maps and sensor fusion, but HD maps are expensive to build and keep current. A positioning layer that delivers continuous, infrastructure-free localization, inside tunnels, indoor and underground space, could reduce dependence on pre-mapped environments and allow vehicles to navigate novel or temporarily altered structures (construction zones, disaster areas) without prior survey.
Juxta's own Nomad micro-store demonstrates the retail robotics angle. The unit uses AI-driven computer vision and advanced shelf sensors to build a digital basket in real time, but its mobility, designed for deployment at EV charging stations, festivals, and campuses, implies a positioning stack that works wherever the unit is dropped. That's a proxy for last-mile delivery robots, security patrol units, and hospital logistics bots: all need to know where they are without installing beacons in every hallway or loading dock.
Healthcare asset tracking, another cited deployment, points to an adjacent robotics use case: mobile manipulators and transport robots in hospitals. These machines move between GPS-available outdoor zones and GPS-denied indoor wards, elevators, and basements. A single positioning modality that bridges both, offline-capable and hardware-free, simplifies fleet management and reduces the sensor suite each robot must carry.
The cost structure matters. By eliminating expensive hardware like receivers, beacons and cameras, UPS targets the bill-of-materials pressure that keeps many robotics pilots from scaling. For a warehouse operator running hundreds of AMRs, removing beacon infrastructure, plus installation and maintenance, shifts the economics. For a defense contractor building unmanned ground vehicles, the anti-jamming claim adds a layer of operational assurance that pure GPS/INS stacks can't provide.
None of this is speculative vaporware. The partnership with TPIsoftware, a Taiwan-based integrator with on-premises AI deployment capability, suggests a path to regional production and support critical for robotics OEMs that need supply-chain certainty and data-sovereignty compliance.
The open question is latency and compute. IMU-plus-ML localization demands onboard inference; TPIsoftware's emphasis on on-premises AI deployment for enhanced data sovereignty hints at edge compute integration. If the stack runs on the robot's existing compute module without a dedicated accelerator, adoption friction drops. If it requires a separate GPU, the cost advantage erodes.
What's clear: the defense-originated technology is already migrating into commercial robotics verticals. The warehouse, the hospital, the tunnel, the festival ground — each is a GPS-denied theater where autonomy has been brittle. A positioning layer that works without satellites, without beacons, and without prior maps changes the deployment calculus. The Juxta–TPIsoftware deal may be the first public signal that this class of infrastructure-free navigation is ready for volume.
Challenges and Open Questions
The promise of satellite-free positioning collides with three hard realities: stitching new terrestrial signals into legacy GNSS architectures, navigating a regulatory environment built around space-based constellations, and paying for infrastructure that doesn't yet exist at scale. Juxta's UPS — built on MEMS IMU sensors and machine learning rather than receivers, beacons, or cameras — sidesteps some hardware costs, but the integration, certification, and deployment questions remain.
Integration: The Layered-Reality Problem
Every AltPNT technology enters a world where GPS is already embedded in avionics, maritime ECDIS, ground-vehicle telematics, and timing networks for finance and telecom. Research on BDS–eLoran fusion shows the technical friction: "Considering the different time delays between the eLoran signal reception channel and the BDS signal reception channel, an additional BDS receiver is required to achieve time synchronization with the eLoran signals." Most existing integration studies "implicitly assume the availability of four or more satellites and primarily focus on performance enhancement under nominal conditions," not the denied environments where Juxta and TPIsoftware are targeting. SpaceNews reporting underscores that "the reality is that we need a layered approach to PNT that interconnects multiple technologies to enable graceful degradation in adverse conditions," and that "these factors make a fully independent system quite difficult, which is why the most advanced tech companies in the world take a layered approach that integrates VPS with other systems." For a defense site or autonomous security fleet, the integration task isn't just adding a sensor; it's rewriting the sensor-fusion stack so that IMU-derived position, visual odometry, and any residual GNSS measurements agree on a common reference frame and failover logic. The PR materials note UPS has been field-tested and deployed across government, defense, logistics, and healthcare environments, but they don't detail the middleware changes those deployments required, a gap that will matter when a prime contractor asks for a standard military or avionics interface.
Regulation: Standards Written for Satellites
Current PNT standards, such as ICAO SARPs for aviation, IMO performance standards for maritime, and ITU-R recommendations for spectrum, assume space-based transmitters. The U.S. "Resilient PNT Reference Architecture" and China's parallel doctrine both emphasize "the ability to adapt to changing circumstances and to respond and recover quickly from disruptions," but neither has codified certification paths for terrestrial-only systems. The General Lighthouse Authorities of the U.K. and Ireland began evaluating eLoran's performance in marine environments in the early 2000s and backed the development of hybrid receivers capable of receiving both GNSS and eLoran signals, yet hybrid-receiver certification remains a national, not international, process. For Juxta's UPS, which operates without dedicated beacons, the regulatory question shifts from spectrum allocation to performance validation: what test regimen proves sub-meter accuracy anywhere on Earth even in circumstances where GPS is denied to a DoD acquisition officer or a port-authority safety auditor? The research is silent on any formal type-acceptance process for IMU/ML positioning, and until one exists, each customer will demand its own validation campaign, a cost and schedule driver no press release addresses.
Cost: Infrastructure vs. Software, and the Mapping Trap
"The most important issue is cost," SpaceNews concluded in its AltPNT survey. LEO constellations require larger constellations to achieve global coverage, more frequent replacements and repairs, and frequent handoffs between satellites that create higher operational costs and complexities. Terrestrial alternatives have their own bill. eLoran requires a heavy infrastructure investment and it only provides lateral positioning, not vertical navigation capability. Visual Positioning Systems — used by companies such as Google, Meta, Apple, Microsoft, Snap, and Niantic — face a mapping trap: in order for this to work, all the pixels generally have to be mapped out ahead of time with highly accurate geospatial coordinates. When operating at a large geographic scale like a city or region, this presents a significant logistical and cost challenge. It also poses unique challenges in a wartime environment; after all, if an area's key landmarks are destroyed by bombs, missiles or fire, the pre-mapped pixels will no longer be accurate. LiDAR typically needs to have the region already mapped so that the device can localize. This can be expensive and also causes limitations since it requires a clear line of sight to function properly (which can be impeded by obstacles and weather conditions) and its effectiveness diminishes over longer ranges. Juxta's pitch, a hardware-free, offline-capable platform that turns any setting into an intelligent, continuously trackable space, aims to bypass beacon and mapping costs, but research on MEMS IMU drift shows error propagates at a scale of approximately 5% of the traveled distance in comparable inertial systems. That drift bound forces periodic re-anchoring — either to a surveyed point, a GNSS fix when available, or a pre-mapped feature — which reintroduces the very infrastructure the system claims to avoid. The PR claim of sub-meter tracking accuracy anywhere on Earth has not been accompanied by a published total-cost-of-ownership model for large-scale defense installations or fleets of autonomous patrol robots.
Open Questions the Partnership Hasn't Answered
- What is the certified mean-time-between-failures for UPS in a contested RF environment where both GPS and terrestrial signals are jammed?
- Does the TPIsoftware integration layer expose a standard NMEA or RTCM stream, or does each prime contractor need a custom API?
- Who owns the liability when an autonomous security vehicle loses position fix inside a hardened facility — Juxta, TPIsoftware, the integrator, or the end user?
- What is the recurring license cost per node, and does it scale sub-linearly for large deployments?
Until those questions have documented answers — not marketing language — the partnership remains a technology demonstration with a reseller agreement, not a program of record.
Outlook: Roadmap and Future Expansions
The partnership announcement makes clear that both companies view this as a market-entry vehicle, not a one-off integration. The immediate target set is Taiwan and the wider APAC region — healthcare, disaster prevention, and logistics, with on-premises AI deployment pitched as the mechanism for data sovereignty, security, and resilience.
That APAC push rests on TPIsoftware's existing footprint. The company, founded in 2005, operates in Vietnam, Singapore, Thailand, and Japan, delivering mission-critical systems for financial, healthcare, government, and manufacturing customers. Its integration layer, built on industry standards, already connects to major back-office and loyalty platforms in the retail-fuel ecosystem via Gilbarco's Passport POS software. Juxta's Universal Positioning System, which fuses IMU sensors with machine learning to deliver sub-meter accuracy without GPS, becomes a new module in that stack.
On the retail-autonomy side, Juxta's Nomad micro-convenience stores provide a parallel expansion track. The first Nomad is open to the public in the United States, and four additional units are in production to fulfill initial orders for retailers in Colorado, Texas, New Mexico, and Georgia. Deployment targets include mixed-use commercial-residential developments, college campuses, music festivals, and EV charging stations distanced from traditional retail amenities. The company has also signaled that traditional gas-station formats and unattended car washes are next in line. Each Nomad runs the same sensor-fusion stack, computer vision plus shelf sensors, that feeds positioning data back into the UPS platform, creating a feedback loop between retail operations and the core navigation technology.
The defense and critical-infrastructure angle remains the least specified in public statements. The announcement notes UPS supports full-scale operation monitoring, continuous positioning inside tunnels, indoor and underground space, and navigation for unmanned systems operating in mountainous or remote areas where signal reception is limited or disrupted. It also highlights resilience against jamming and spoofing, vulnerabilities inherent to satellite-dependent GPS. But no DoD program office, prime-contractor teaming agreement, or specific solicitation has been named. The language stops at field-tested and deployed across government, defense without attributing a contract vehicle or funding line. That silence is notable: in this sector, a named program usually appears in a press release if it exists. Its absence suggests the partnership is still in the pre-program phase, building the technical and compliance case before pursuing formal acquisition pathways.
Funding details for either company are not disclosed in the available materials. Juxta operates as a venture of Gilbarco Veeder Root and Vontier Corp., a structure that typically means capital comes from the parent's balance sheet rather than external rounds. TPIsoftware, a private Taiwanese enterprise-software house, has not announced a raise. The roadmap therefore hinges on revenue traction from the reselling network TPIsoftware will manage and the system-integration services it will provide across multiple regions. If the APAC deployments in healthcare and disaster prevention convert to recurring revenue, that cash flow becomes the de facto funding mechanism for further UPS hardening, potentially including the radiation-tolerant, vacuum-qualified variants a space-adjacent application would demand. Until a customer or agency commits to that qualification campaign, space-related use cases remain speculative.
The next concrete milestones are visible: the four Nomad stores shipping to U.S. retailers, the integration of UPS into TPIsoftware's on-premises AI solutions for Taiwanese and APAC healthcare and logistics customers, and the expansion of the reselling network across Vietnam, Singapore, Thailand, and Japan. Whether that momentum translates into a named defense program of record — or a space-qualified derivative — will depend on whether the anti-jamming performance demonstrated in tunnels and mountains survives the scrutiny of a formal test-and-evaluation event. The partners have built the technical case; the procurement case is still being written. The first prime contractor to sign an integration contract will write the next line.
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.