Teardown

Construction · Deep dive

Gecko Robotics

Wall-climbing robots for Navy ships, power plants and refineries — trying to turn inspection data into the system of record for critical-asset health.

emerging

The question that decides it: Can robots-plus-Cantilever displace the manual NDT ecosystem (NDT technicians, insurance surveyors, class societies) fast enough to make Cantilever the system of record for critical-asset health — before Palantir Foundry, Cognite or a Boston Dynamics / Shield AI acquirer bundles inspection data into their own industrial platform and reduces Cantilever to a sensor feed?

My take

Founded
2013
Revenue
Undisclosed; lifetime cumulative <$71M implied through Mar 2026 (Pulse, Mar 2026)
Screen
Published
2026-11-18
Web
www.geckorobotics.com
Elsewhere
LinkedIn · Crunchbase

Founders and leadership

  • Jake Loosararian Co-founder & CEO

    Grove City College electrical engineering (2013); built the first Gecko prototype as a college project after visiting a nuclear-plant boiler inspection; rejected a $500K acquisition offer in 2015; YC Winter 2016 cohort.

  • Troy Demmer Co-founder & Chief Product Officer

    Grove City College classmate of Loosararian; joined full-time after Gecko entered YC Winter 2016; CERAWeek speaker on robotics in energy.

Snapshot

Gecko Robotics sends magnetic wall-climbing robots into power plants, refineries, Navy ships, and pipelines to collect millions of ultrasonic thickness readings per hour, then feeds the data into Cantilever, its proprietary AI asset-health platform. The Pittsburgh company closed a $125M Series D led by Cox Enterprises at a $1.25B valuation in June 2025 (Gecko, Jun 2025); by October 2026 its U.S. Navy IDIQ ceiling had quadrupled from $71M to $300M in seven months (Maritime Executive, Oct 2026). The thesis: physical inspection data is the missing raw input for the industrial AI stack, and whoever captures it owns the system of record for critical-asset health. The real question is whether a hardware-heavy company can win the data-platform fight before Palantir-class software incumbents bolt on robot ingestion.

Founding story

Jake Loosararian built the first Gecko prototype as an electrical-engineering student at Grove City College in 2012 after visiting a local coal plant and watching inspectors work on ropes inside a boiler (Y Combinator blog, 2016). The 40-pound climber he and classmates built carried an ultrasonic probe and reportedly found cracks a rope-access team had missed. By 2015 he had burned through savings and was offered $500K to sell (CNBC/NBC, Feb 2025). He refused, got into YC's Winter 2016 cohort, and brought in co-founder Troy Demmer, a Grove City classmate who became CPO. The pair's bet was that the ~$4B spent annually on manual industrial NDT inspections (Sacra, 2025) was mostly collecting bad data — spot ultrasonic grids that covered maybe 1% of an asset — and that a robot capable of full-wall coverage was the shortest path in.

How it works

Field teams of one or two operators carry a TOKA-series robot into the asset — TOKA 3 for medium piping, TOKA 4 for curved boiler walls, TOKA 4 GZ for tank floors, TOKA Flex for small-diameter piping, MONARCH for confined spaces (Contrary Research, 2024). Magnets or vacuum attachment let the robot cling to vertical steel and crawl at 9–18 m/min. Onboard arrays of 8–96 ultrasonic transducers take thickness readings every few millimeters alongside 1080p video. A typical pass covers ~600 sq ft per 12-hour shift, versus the ~1% asset coverage a rope-access ultrasonic-gridding team delivers (Sacra, 2025).

Raw readings stream through Fulcrum, Gecko's data pipeline, into Cantilever, the cloud platform launched October 31, 2023 (Robot Report, Oct 2023). Cantilever fuses the robot pass with historical records, design files, SCADA telemetry, and operational data, then produces 3D color-coded thickness maps, corrosion rates, remaining-life estimates, and prioritized repair lists — typically within 24–48 hours. Navy repair plans that previously took weeks are reportedly now generated in minutes (Robotics 247, Oct 2023).

Product and business overview

Gecko sells three bundled things. The hardware fleet — the TOKA family plus the MONARCH confined-space robot — is proprietary and never sold outright; it travels with field crews. The field-service layer bills by scope: a mapped boiler, a ship's ballast tank, a refinery piping run. The Cantilever software subscription gives customers a persistent 3D model of each asset with corrosion deltas tracked over time (Gecko, 2024). On top of those three, a growing engineering-services line sells repair design, digital twins, and class-society compliance reports (Sacra, 2025). Industry coverage spans six verticals: power (coal, gas, nuclear, hydro), Navy surface fleet and submarines, oil & gas, manufacturing, pipelines, and increasingly new-build shipyard QA for Columbia-class submarine production (iConnect, 2025).

Business model and pricing

Gecko is a robotics-as-a-service business with a software annuity on top — not a hardware vendor. Per the Y Combinator blog, inspection services run roughly $50K–$100K per robot deployment (YC, 2016); multi-year contracts tie Cantilever access, data storage, and analytics to the recurring scans. Pricing is structured by asset scope and frequency rather than hourly rates (Sacra, 2025). The company has shown it can size deals few inspection competitors can: a $300M Navy IDIQ for 18+ ships (Maritime Executive, Oct 2026), a $100M strategic deal with NAES to modernize U.S. power plants (Robot Report, Feb 2025), and three simultaneous ADNOC Gas contracts signed at ADIPEC 2025 (The National, Nov 2025). Sacra estimates a single deployed TOKA generates hundreds of thousands of dollars in annual revenue (Sacra, 2025).

Traction over time

Gecko has never disclosed revenue, but trade coverage is suggestive. When the Navy IDIQ was first awarded in March 2026 at a $71M ceiling, Pulse's analysis described it as "eclipsing lifetime revenue" (Pulse, Mar 2026), implying cumulative sales through 2025 were comfortably under $71M. By late 2026, the ceiling had quadrupled.

Date Milestone Source
2013 Company founded in Pittsburgh GCC, 2016
Winter 2016 YC cohort, Mark Cuban seed YC, 2016
Oct 2020 BPX Energy case study BP, Oct 2020
Mar 2022 $73M Series C at $533M post-money (XN) Sacra, 2025
Oct 2023 Cantilever platform launched Robot Report, Oct 2023
Dec 2023 $100M Series C extension at $633M Sacra, 2025
Feb 2025 $100M NAES partnership T&D World, Feb 2025
Jun 2025 $125M Series D at $1.25B led by Cox Gecko, Jun 2025
Nov 2025 Three ADNOC Gas contracts at ADIPEC The National, Nov 2025
Mar 2026 Navy IDIQ awarded, $71M ceiling / $54M initial Gecko, Mar 2026
Oct 2026 Navy IDIQ raised to $300M Maritime Executive, Oct 2026

Headcount sat at roughly 266 on Unify GTM's most recent snapshot, up ~7% year over year (Unify GTM, Nov 2025); LinkedIn-derived estimates put late-2026 count closer to 300. Loosararian told Fortune in early 2025 he was still running "100-hour weeks" through the Series D (Fortune, Feb 2025).

Market analysis

The markets Gecko sells into are large and growing. StraitsResearch pegs the NDT robotic-inspection segment at $44.1B by 2032, growing in the mid-teens annually (StraitsResearch, 2026); the broader industrial NDT services pot is roughly $20B today. DoD depot maintenance alone will spend over $130B over the next five fiscal years (Sacra, 2025), and unplanned downtime across ~50,000 industrial plants costs an estimated $50B a year. The structural tailwinds are favorable: U.S. utilities are extending 40-year-old fossil assets to meet AI data-center demand; the Navy's surface fleet sits at ~60% readiness vs. an 80% target (Washington Examiner, Oct 2026); and Gulf NOCs chase corrosion losses reportedly above $10B annually.

Competitive intel

Competitor Scale / angle of attack
Boston Dynamics (Spot) Hyundai-owned, ~$1B+ revenue run-rate. Quadruped platform with a rich partner ecosystem; hardware reportedly ~2x Gecko's cost, and the stack is general-purpose rather than NDT-tuned (Sacra, 2025).
ANYbotics Zurich-based, $50M+ raised. ANYmal quadruped targets oil-and-gas inspection; relies on partners (incl. GE Vernova) for NDT payloads (CB Insights).
Flyability Confined-space drones (Elios 3). Strong in boiler and tank visual inspection; weak on ultrasonic thickness coverage. Directly addresses the segment MONARCH targets.
MISTRAS / Applus+ / SGS Legacy TIC / NDT service providers, each with $1B+ revenue and incumbent utility and refinery contracts. Deep technician networks; no proprietary robot stack, no comparable software layer.
Cognite / Palantir Foundry Industrial data-ops incumbents. Neither captures inspection data today, but both can ingest it. The real threat is bundling: if Foundry or Cognite becomes the asset-health system-of-record, Cantilever is reduced to a sensor feed (LNS Research, 2025).
GE Vernova / Honeywell Forge / Siemens Energy OEM enterprise platforms running their own inspection divisions on partner hardware. They can protect existing service contracts by "making inspection look like a feature, not a product" (Sacra, 2025).

History and evolution

The arc has four acts. 2013–2016: college prototype, near-bankruptcy, YC, a rejected $500K acquisition offer. 2017–2021: TOKA family build-out, early marquee customers (Dominion, Duke, BP, ExxonMobil), Mark Cuban / Founders Fund / Drive Capital on the cap table. 2022–2023: Siemens Energy global partnership (Oct 2022), $73M Series C led by XN (Mar 2022), Air Force Nuclear Weapons Center contract (Nov 2022), the Cantilever launch (Oct 2023), and a $100M Series C extension at $633M (Dec 2023). 2024–2026: the defense pivot. The Cox-led $125M Series D (Jun 2025) was framed around "building and modernizing" infrastructure rather than inspection per se (Gecko, Jun 2025); within eight months Gecko landed the Navy IDIQ (Mar 2026) and quadrupled it (Oct 2026). Publicly admitted failures are confined to the lossy pre-YC years; the strategy-churn and management-experience themes Glassdoor flags have not been externally reported.

What people say

The case for. Trade press is overwhelmingly favorable. Fortune ran a founder-archetype profile on Loosararian (Fortune, Feb 2025); Breaking Defense and Maritime Executive covered the Navy expansion as a legitimate readiness lever, with reported claims of 83% faster repair-plan turnaround and "tens of millions saved per vessel" (Washington Examiner, Oct 2026). Sacra argues the vertical integration — robots tuned to Cantilever — is a structural moat over ANYbotics-plus-partner stacks (Sacra, 2025). Internal culture shows 76% CEO approval on Glassdoor, with reviewers citing mission, autonomy, pay, and the Navy traction.

The complaints. Glassdoor overall is 3.4/5 on 110 reviews — below the 3.8 IT industry average — and only 60% would recommend (Glassdoor, Nov 2026). Recurring themes: "spread thin" across too many industries, frequent change of direction, managers lacking experience, and one review titled "Corporate Cult" flagging high-pressure mission-rhetoric culture. Pre-IPO coverage is less bullish: Pulse's observation that the initial $71M Navy ceiling "eclipsed lifetime revenue" (Pulse, Mar 2026) is effectively a short line — nine years of commercial sales to under $71M cumulative. Sacra's platform-bundling note is harsher: Honeywell, GE, and Siemens can "make inspection look like a feature, not a product" (Sacra, 2025).

Outlook: the open question

The open question is whether Gecko can turn Cantilever into the system of record for critical-asset health before a software incumbent — Palantir Foundry, Cognite, or a Boston Dynamics / Shield AI acquirer — makes robot-captured inspection data a cheap ingestion feature inside a broader industrial platform. Gecko wins if two things happen: Cantilever opens to third-party sensor and robot feeds (so a utility doesn't have to buy Gecko hardware to get the data layer), and the Navy deal cements Cantilever as the default asset-health record for the ~$130B DoD depot maintenance pool (Sacra, 2025). It loses if Palantir ships a Foundry template for asset-health that any existing TIC service provider can pipe data into first — because inspection fieldwork is lower-margin and more easily commoditized than the software layer above it. The nine-months-to-quadrupled-Navy-ceiling arc suggests the first path is live; the fact that no commercial revenue number has been disclosed through all of 2026 suggests the industrial expansion is slower than the defense story.

How to attack it

The wedge is a vendor-neutral Cantilever. Build a software-first asset-health platform that ingests inspection data from any source — Flyability drones, ANYmal quadrupeds, incumbent MISTRAS and Applus+ ultrasonic reports, even phone-camera photos from pipeline walkdowns — and compete on analytics rather than capture. Gecko cannot easily respond because opening Cantilever to competitor hardware cannibalizes the robots-as-a-service annuity that funds R&D; Gecko's own Sacra thesis positions the vertical stack as the moat (Sacra, 2025).

The exploitable weaknesses are concrete. Customer concentration: growth leans on one Navy IDIQ and a handful of marquee utility accounts; the Pulse "eclipses lifetime revenue" line implies a thin commercial base (Pulse, Mar 2026). A competitor offering utilities a cheaper software layer that doesn't require Gecko crews gains ground quickly. Cost structure: field teams, custom hardware, and ~266 employees carry high fixed costs (Unify GTM, Nov 2025); a software-only competitor has roughly a 10x gross-margin advantage. Channel dependency: big deals (NAES, L3Harris, ADNOC via Al Masaood Energy) flow through partner channels; a partner squeeze squeezes Gecko. Cultural debt: "corporate cult" and "spread thin" themes on Glassdoor (Glassdoor, Nov 2026) signal the engineering org may struggle to open Cantilever cleanly to third-party data. Hardware debt: the TOKA family is magnetic-coupled, which locks Gecko out of non-ferrous assets — concrete bridges, FRP tanks, GRE pipe in Gulf chemical plants — where legacy NDT still holds share. An attacker starting with drones and phone-camera photos on non-magnetic infrastructure has a clean lane.

Adjacent-segment play

The same robot-plus-analysis stack maps naturally onto civil infrastructure — bridges, dams, tunnels, water-treatment plants — where federal spend ($1.2T IIJA) is live and asset inventories are even more neglected than power plants. ASCE rates most U.S. bridges "C" or worse; a magnetic climber can't scale concrete, but a drone-plus-LiDAR-plus-GPR variant fed into Cantilever could. Gecko has flirted with this via rhetoric ("D-grade infrastructure") but hasn't built the hardware.

A second adjacency is building envelope — façade inspection, roofing, solar-panel soiling — partly colonized by Zeitview (formerly DroneBase) and DroneDeploy. These are software-heavy, hardware-light businesses growing against insurance-driven inspection demand; Gecko is unlikely to pivot there because the cultural and hardware mismatch is severe and margins look lower than defense NDT. More realistic is maintenance orchestration — repair crews, parts procurement, work orders — where Sacra notes Gecko is already pushing (Sacra, 2025). That lands Gecko where CMMS vendors like UpKeep and Fiix (Rockwell) operate but anchored by first-party inspection data — a credible repackage of the stack, with execution risk.

Sources and further reading

Capital history

DateRoundAmountValuationLead(s)
2016 YC W16 / Seed Undisclosed — —
2018-2020 Series A/B (multiple tranches) ~$40M cumulative — —
Mar 2022 Series C $73M $533M post-money —
Dec 2023 Series C extension $100M $633M post-money —
Jun 2025 Series D $125M $1.25B post-money —

Investors / owners: Cox Enterprises, US Innovative Technology Fund (USIT), XN, Founders Fund, Y Combinator, Drive Capital, Mark Cuban, Joe Lonsdale, Snowpoint Ventures, Gokul Rajaram, XYZ Ventures

Competitive set

  • Boston Dynamics (Spot) — Hyundai-owned quadruped platform; hardware reportedly ~2x Gecko's cost, general-purpose rather than NDT-tuned.
  • ANYbotics — Zurich quadruped firm; ANYmal sold to O&G majors through partners like GE Vernova that supply NDT payloads.
  • Flyability — Elios 3 confined-space drone; visual-inspection strong, limited ultrasonic coverage — directly targets Gecko's MONARCH use cases.
  • MISTRAS / Applus+ / SGS — Legacy TIC/NDT service providers with $1B+ revenue each and entrenched utility/refinery contracts; large technician networks but no proprietary robot stack.
  • Cognite / Palantir Foundry — Industrial data-ops platforms that can ingest Gecko-style data; the bundling threat is turning Cantilever into a feed inside their asset-health workflow.
  • GE Vernova / Honeywell Forge / Siemens Energy — OEM enterprise platforms with their own inspection divisions; can 'make inspection look like a feature' to defend larger software contracts.