Logistics / Autonomous trucking · Deep dive

Gatik

The autonomous-trucking company that skipped the glamorous robotaxi race and the long-haul highway war to master the boring, repeatable middle mile — driver-out box trucks looping between Walmart and Kroger depots — betting that a constrained problem is the only one that pays.

emerging

The question that decides it: Gatik's whole thesis is that deliberately shrinking the problem — Level 4 autonomy confined to fixed, repeatable, sub-highway middle-mile routes ('structured autonomy') — lets it strip out the safety driver years before long-haul rivals and turn each truck into a positive-margin unit under a five-year, take-or-pay contract. Does that per-truck economics actually clear once Gatik scales from ~100 driverless trucks (2026) into the 'hundreds' it has promised — absorbing the capital cost of fleet expansion, the 2027 Isuzu purpose-built truck transition and remote-supervision overhead — before its customer concentration (Walmart and a handful of Fortune 50 retailers) or the sector's chronic timeline slippage catches up with it?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Gatik is a Mountain View, California autonomous-trucking company that made a deliberately unglamorous bet: skip the robotaxi race and the long-haul highway war, and automate the “middle mile” — short, repeatable B2B runs between distribution centers and stores. Founded in 2017 by two Carnegie Mellon-trained brothers and a third robotics engineer, it runs Level 4 driver-out box trucks on fixed routes for Fortune 50 customers including Walmart, Kroger and Tyson Foods. In January 2026 it declared itself the first U.S. company operating fully driverless trucks at commercial scale, citing more than $600 million in contracted revenue and ~60,000 driverless orders since mid-2025. It has raised more than $200 million (CEO, May 2024; trackers put it higher), including a strategic investment from Isuzu that will mass-produce purpose-built trucks from 2027. The thesis — a smaller problem is the only profitable one — is elegant; whether it survives fleet-scale capital costs and Walmart-heavy concentration is the open question.

Founding story

Gatik exists because its founders looked at the mid-2010s AV gold rush and concluded almost everyone was solving the wrong problem. Brothers Gautam Narang (CEO) and Arjun Narang (CTO), both trained in robotics/CS at Carnegie Mellon, had spent a decade on autonomous systems together — including outdoor security robots built with third co-founder Apeksha Kumavat (Chief Engineer, MSECE Purdue 2016).

Their contrarian insight, formed at the 2017 founding: robotaxis face an unbounded set of situations and long-haul trucking still spans thousands of unpredictable miles, but the middle mile — the same short loop between warehouse and store, over and over — is a problem you can constrain. Gatik calls this “structured autonomy”: it solves a specific, mapped, repeatable route and refuses to operate outside it. That constraint, Gautam Narang argues, is what let Gatik remove the safety driver years before long-haul players and book revenue doing it. Most AV founders sold a moonshot; Gatik sold a delivery service that happens to be autonomous.

How it works

When a Gatik truck runs, it executes a route it has driven hundreds of times. Gatik operates Class 6/7 medium-duty box trucks — 26- and 30-foot vehicles, many modified Isuzus — loaded at a distribution center or “dark store” and driven to a nearby retail location and back. The flagship was a 7.1-mile loop in Bentonville, Arkansas, run daily without a human aboard from 2021.

The mechanically important choice is the tightly bounded operational design domain. Routes run largely on city and suburban surface streets at up to ~45 mph (highway running up to ~65 mph has begun, with a safety driver), on densely mapped, validated roads. Refusing the open world shrinks the edge-case long tail that has sunk other AV programs — which is why Gatik could pull the safety driver far earlier than long-haul peers. Redundant sensing (lidar, radar, cameras), a purpose-built compute stack and remote human supervision back up on-vehicle autonomy. In 2025 Gatik standardized on Nvidia’s DRIVE AGX (Thor SoC) for its next-gen trucks. By January 2026 it ran trucks nearly around the clock across Texas, Arizona, Arkansas and Ontario, moving ambient, refrigerated and frozen goods.

Product and business overview

Gatik does not sell trucks or software; it sells a driverless delivery service on a route. A customer hands Gatik a recurring short-haul lane it already runs, and Gatik operates it autonomously under a multi-year contract, absorbing the vehicles, autonomy, mapping and supervision. Named customers span Walmart (the anchor), Kroger, Tyson Foods, Georgia-Pacific/KBX into Sam’s Club stores, Loblaw in Ontario, and Pitney Bowes. The pitch is prosaic and durable: more frequent replenishment at lower cost, without the chronic driver shortage — not a robotaxi, but shelves that stay stocked.

The strategic capstone is the Isuzu partnership: rather than retrofit trucks forever, the two agreed in 2024 to mass-produce purpose-built L4 trucks at a new Isuzu factory from 2027, on Nvidia compute, with capacity reportedly scaling toward ~50,000 vehicles a year by 2030 — the difference between a science project and an industrialized fleet operator.

Business model and pricing

Gatik’s model is transportation-as-a-service on founder-favorable terms. Customers sign roughly five-year agreements with no termination-for-convenience clause, so Gatik carries no utilization risk — it gets paid whether or not a truck is fully loaded. It leases the vehicles (from Ryder and Holman) and folds the lease cost into the per-route fee. Per-mile prices are undisclosed, so external unit economics stay opaque; what Gatik markets is the revenue quality — long-dated, contracted, take-or-pay lanes, framed at more than $600 million cumulative by January 2026.

That contract structure is the quiet strength of the story. In an industry where “revenue is still not the highlight” (CEO, May 2024), five-year no-out contracts let Gatik argue for “profitable growth” rather than perpetual burn, with profitability targeted “later this decade.” The unproven part is per-truck margin at scale: contracted revenue is a top-line number, not evidence that each driverless truck clears a positive contribution margin once the fleet runs into the hundreds.

Traction over time

The trajectory is real but the headline metrics deserve scrutiny. The striking numbers — $600M contracted revenue, 60,000 orders — are self-reported; the 10,000+ cumulative driverless miles (early 2026) is modest next to Aurora’s 12 million-plus, and reflects activity, not disclosed revenue or margin. Every figure here is a leading indicator of scaling, not proof of profitability.

Market analysis

Gatik sizes its own opportunity — the North American middle mile — at roughly $250 billion (CEO, 2024), a figure capturing all short-haul freight it could theoretically automate rather than the autonomous slice. That slice is smaller and more speculative: Future Market Insights pegged the middle-mile autonomous delivery (B2B road) market at ~$490 million in 2026, growing ~40% CAGR to ~$14.2 billion by 2036, with L4 box trucks and transport-as-a-service leading.

The structural tailwinds are genuine: a persistent driver shortage, high labor cost, retailers’ hunger for more frequent replenishment, and the fact that fixed repetitive routes are the most automatable freight task. The countervailing force is the sector’s credibility problem — autonomous trucking has burned through capital and timelines for a decade, and driver-out freight has repeatedly proven smaller and later than promised. Gatik’s constrained-ODD strategy answers that history, but also caps its near-term TAM to lanes short and repeatable enough to qualify.

Competitive intel

The competitive set splits cleanly by geography of the road (full profiles in the sidebar). The scaled players — Aurora Innovation (Nasdaq: AUR, 12M-plus autonomous miles, driver-out Dallas-Houston long-haul) and Kodiak AI (public via a ~$2.5B SPAC in September 2025, Permian and Texas long-haul) — are both bigger, better-capitalized and public, but both chase long-haul highway freight, a larger and different TAM than Gatik’s middle mile. Waabi and Torc Robotics (Daimler-owned, 2027 launch) round out the highway camp; Plus competes for the same OEM deals. Closest to Gatik’s mission is Einride, whose cab-less electric Pods target short/middle-mile freight.

Where Gatik wins: it is arguably furthest along on true driver-out commercial operations (having removed the hardest variable, open-world highway driving), has recurring contracted revenue rather than pilots, and the Isuzu deal gives it a path to industrialized trucks most rivals lack. Where it is exposed: the long-haul players command far more capital and a bigger prize, its route TAM is narrower, and its “first driverless at scale” claim rests on short loops and modest mileage. The backdrop is the graveyard: TuSimple wound down U.S. operations in late 2023; Embark collapsed in 2023 (absorbed by Applied Intuition); Waymo Via was shelved in July 2023; Nuro cut staff. Gatik’s differentiation is that it is not on that list.

History and evolution

• 2017 — Founded by Gautam Narang, Arjun Narang and Apeksha Kumavat; middle-mile-only from inception.
• Nov 2020 — $25M Series A co-led by Wittington Ventures and Innovation Endeavors.
• Aug-Nov 2021 — Removes the safety driver on the Bentonville, Arkansas loop for Walmart — a worldwide first; $85M Series B led by Koch Disruptive Technologies; expands to Texas.
• 2022-2023 — Adds Georgia-Pacific/KBX (Sam’s Club), Tyson, Kroger and Loblaw (Ontario); ~$85M Series C at a reported ~$700M valuation.
• May-Oct 2024 — Isuzu invests $30M and commits to mass-produce L4 trucks from 2027 (total raised passes $200M); Nippon Express adds a strategic round; completes an independent safety-case conformity assessment.
• Jan 2026 — Declares itself the first U.S. company operating fully driverless trucks at commercial scale; $600M contracted revenue; goal of “hundreds” of trucks by year-end.

No public collapse — but the hard chapters (proving per-truck margin, funding a hundreds-of-trucks fleet, executing the 2027 Isuzu transition, and surviving investor patience in a scarred sector) are ahead, not behind.

What people say

The case for. Trade press treats Gatik as the pragmatist that got autonomous freight to actually work: Forbes has framed its middle-mile focus as “the path to profitable AVs.” The five-year take-or-pay contracts and marquee logos (Walmart, Kroger, Tyson, Loblaw) are strong validation, as is Isuzu’s willingness — a first for a truck OEM — to commit a factory to mass-producing L4 trucks, which the CEO calls validation of “the technology, the use case and the economics.” Nvidia and Goodyear partnerships and a 2024 independent safety-case assessment add credibility. In a sector littered with failures, Gatik’s driver-out record and contracted revenue are differentiators peers cannot show.

The complaints. The bear case is partly cultural, partly structural. Glassdoor reviews describe a burnout culture, founder micromanagement, thin HR support, and — most alarming for a safety-critical company — claims that a large share of engineers, including senior members of the autonomy stack, left over a compressed period, plus grievances about COVID-era in-office mandates. Safety-stack churn is the soft signal that precedes hard AV problems. Structurally: customer concentration is real (Walmart anchors the story); the marquee metrics are self-reported and flatter short loops; the total-funding figure is muddled across trackers ($200M+ per the CEO, up to ~$330M elsewhere), signaling opacity; and the sector’s history — TuSimple, Embark, Waymo Via, Nuro — warns that timelines slip and capital runs out. Profitability remains “later this decade” — unproven.

Outlook: the open question

Gatik works if “structured autonomy” on fixed middle-mile routes produces a genuinely positive per-truck contribution margin — net of vehicle, remote supervision, mapping and safety operations — and that margin holds as the fleet scales from ~100 driverless trucks into the promised hundreds, funded through the capital-intensive 2027 Isuzu transition without a dilutive scramble. It fails if the economics only pencil at demo scale, if fleet expansion outruns the balance sheet, if Walmart-heavy concentration leaves it hostage to one buyer’s roadmap, or if the sector’s timeline slippage and the safety-stack churn seen in reviews surface as setbacks before profitability arrives. The bet is narrower — and more falsifiable — than most AV bets.

The bull case is coherent and, unusually for this sector, grounded in revenue: Gatik chose the one autonomy problem that is actually constrained, pulled the driver years ahead of rivals, locked in utilization-risk-free five-year contracts, and secured an OEM willing to industrialize the truck. The bear case is equally concrete: the slice is narrow, the metrics flatter short loops, the customer base is thin, the funding picture opaque, and the AV-trucking graveyard is full of companies that also had marquee logos. What settles it for Gatik: credibly-signaled positive per-truck unit economics, a fleet that reaches the hundreds on schedule, diversification beyond Walmart, and a clean safety record through scale-up. What settles it against: a fleet that stalls because each truck loses money, a punitive capital raise, a safety or regulatory incident, or the discovery that “driverless at scale” was a demo dressed as a business. The next 18-24 months decide which story this is.

How a challenger would attack it

Sell the truck, not the service. Gatik’s transportation-as-a-service model absorbs vehicles, leases, mapping, and remote supervision onto its own P&L — which is why per-truck margin at scale is the unresolved question and why it needs the 2027 Isuzu factory to pencil. A challenger flips to the Kodiak structure: sell the autonomy stack on customer-owned trucks, letting Walmart-scale retailers — who already run private fleets — own the capital cost while the challenger books software margins. That attacks Gatik where its five-year take-or-pay contracts can’t protect it: at renewal, when a Fortune 50 procurement team compares paying Gatik’s bundled per-route fee against licensing autonomy for its own fleet. The second vector is talent: Glassdoor accounts of burnout, founder micromanagement, and senior autonomy-stack departures mean the engineers who built structured autonomy are hireable, and the constrained-ODD playbook — fixed routes, dense mapping, sub-45-mph surface streets — is a strategy, not a patent; it is the most replicable architecture in the AV sector precisely because it avoids the open-world moat problem. Third, exploit concentration: Gatik’s story is anchored on Walmart and a handful of retailers, so a challenger that signs the number-two player in each vertical Gatik hasn’t locked — pharmacy, convenience, foodservice distribution — boxes it into its existing logos while the fleet-scale capital question is still open.

Same playbook, new buyer

Structured autonomy on fixed loops transfers cleanly to buyers Gatik’s retail-replenishment focus ignores. The nearest adjacency is closed-campus and port freight: container drayage between terminal and transload, airport cargo aprons, and plant-to-warehouse shuttles at manufacturing campuses are even more constrained than Bentonville surface streets — private roads, no NHTSA-exposure debate, faster regulatory path — and the buyer (terminal operators, manufacturers) is desperate for the same driver-shortage relief. Second: middle-mile for grocery and foodservice distributors in Canada and Japan — Gatik’s own investors point there (Loblaw in Ontario is live; Isuzu and Nippon Express are Japanese), but its operational center of gravity is US Sunbelt retail, leaving Japan’s acute driver shortage (the “2024 problem”) open for a partner-led local operator before Gatik’s 2027 truck arrives. Third: parcel and mail linehaul between sortation centers — Pitney Bowes hints at it, but USPS-style contract linehaul runs identical fixed loops at national scale and no AV player owns it. Gatik can’t chase all of these: its capital is committed to reaching “hundreds” of trucks for existing retail contracts, and every new vertical means new mapping, new safety cases, and new supervision infrastructure its balance sheet can’t parallelize.

Sources and further reading

• Startup Gatik Lands Partnership – And $30 Million – From Isuzu To Scale Up Robot Trucks (Forbes, May 14 2024)
• Forget Robotaxis. Upstart Gatik Sees Middle-Mile Deliveries As The Path To Profitable AVs (Forbes, September 5 2023)
• Gatik and Walmart Achieve Fully Driverless Deliveries in a First for the Autonomous Trucking Industry Worldwide (BusinessWire, November 8 2021)
• Gatik Becomes First U.S. Company to Operate Fully Driverless Trucks at Scale for Commercial Deliveries (BusinessWire, January 27 2026)
• ‘Hundreds’ Of Gatik Robot Delivery Trucks Headed For U.S. Roads (Forbes, January 28 2026)
• Gatik Raises $25 Million in Series A Funding (Automotive Fleet, November 23 2020)
• Gatik expands autonomous box truck operations to Texas with $85 million in new funds (TechCrunch, August 31 2021)
• Gatik and Isuzu Team Up to Mass-Produce SAE Level 4 Autonomous Trucks Powered by NVIDIA (Gatik, March 2025)
• The State of Autonomous Trucking in 2025: A Recap of 2024 (NACFE, 2025)
• Autonomous Truck Developer Kodiak AI Goes Public (Transport Topics, September 2025)