Logistics · Deep dive
Kodiak AI
Autonomous-trucking developer; eight driverless sand-haul trucks live in the Permian for Atlas; de-SPAC'd Sep 2025 at ~$2.5B EV.
emerging
The question that decides it: Does Kodiak reach driverless on-highway commercial revenue in H2 2026 before its ~$275M post-SPAC cash reserve is spent, and does it close the 18-month commercialization gap to Aurora?
My take
- Headcount
- ~300
- Screen
- Published
- 2026-11-20
- Web
- kodiak.ai
- Elsewhere
- LinkedIn · Crunchbase
Founders and leadership
-
Don Burnette Founder, CEO
Ex-Google/Waymo (2009-2016); co-founded Otto with Anthony Levandowski (acquired by Uber ATG for $680M); 16+ years in self-driving.
-
Paz Eshel Co-founder (departed 2021)
Former investor at Battery Ventures; left to rejoin Battery.
Snapshot
Kodiak AI is one of three surviving autonomous-trucking developers, alongside Aurora Innovation and Waabi, and the only one focused initially on off-highway haul work rather than public-interstate freight. The Mountain View-based company, founded 2018 by Don Burnette (ex-Waymo, ex-Otto, ex-Uber ATG), went public in September 2025 through a merger with Ares Acquisition Corporation II (NASDAQ: AACT → KODA) at a ~$2.5B enterprise valuation, netting ~$275M of fresh capital including $145M of PIPE and $62.9M of trust cash (FreightOwner, Oct 2025; FreightWaves, Sep 2025). Its first commercial deployment runs eight driverless trucks for Atlas Energy Solutions in the Permian Basin, hauling frac sand across 75,000 square miles with no human safety driver.
Founding story
Don Burnette started on self-driving at Google's X in 2009, moved to Waymo, then co-founded Otto with Anthony Levandowski in 2016 — the company that Uber bought for $680M and whose founder became the subject of the Waymo-vs-Uber trade-secret litigation. Burnette left in 2018 and immediately founded Kodiak Robotics with Paz Eshel (former investor at Battery Ventures), choosing autonomous trucks specifically rather than robotaxis because he had convinced himself that highway-to-highway freight was a simpler problem than city driving but with a market multiple larger than Waymo's TAM. He is the lead technical voice on Kodiak's "Kodiak Driver" architecture, which emphasizes a modular sensor + compute stack that can be retrofitted onto third-party trucks. As of October 2025 Burnette has 16+ years in self-driving — longer than almost anyone running an AV company today (FreightWaves, Oct 2025).
How it works
Kodiak's technology is called the Kodiak Driver — a sensor suite (lidar, radar, cameras), a compute stack, and an autonomy software layer that is bolted onto an OEM truck via a partnership with Roush, the Michigan vehicle engineering shop. Roush does the base-truck modifications (steering, brake, throttle redundancy) and installs the Kodiak kit; the truck ships ready to roll. The architecture is explicitly modular — same software can run on multiple truck makes — which is the counter-approach to Aurora's deeper OEM integration with Paccar and Volvo. In commercial operation at Atlas's Permian site, trucks operate without an observer, running fixed and semi-fixed routes between load-out and well-pads across 75,000 square miles of West Texas (FreightWaves, Oct 2025). For highway freight, Kodiak uses a transfer-hub model: human drivers do first- and last-mile in city, autonomous trucks do the long-haul interstate, and freight is swapped at designated terminals outside major metros.
Product and business overview
Three product lines. Kodiak Driver for Off-Highway — the commercially deployed product, currently with Atlas Energy Solutions in the Permian. Target markets: oil and gas sand hauling, mineral and metal mining, forestry logging roads. These are closed-site operations where regulatory barriers are lower and the customer owns the acreage. Kodiak Driver for On-Highway — the long-haul interstate freight product, in commercial pilots with J.B. Hunt, Werner Enterprises, and Bison Transport. Target for driverless on-highway: H2 2026 (Kodiak, Oct 2025). Driver-as-a-Service pricing model — Kodiak sells autonomy per-mile or per-truck-month rather than selling vehicles outright, which lets fleets keep their existing truck financing and maintenance arrangements.
Business model and pricing
Kodiak's commercial model is Driver-as-a-Service. For on-highway freight, Kodiak's framing is cost per mile — the mix of autonomous hardware/software subscription, insurance, and remote operations fees — benchmarked against a driver-operated baseline of roughly $1.70-1.90/mile all-in (ATRI, 2024). Kodiak has said its per-mile economics target 10-20% below the manned baseline at full utilization. For off-highway (Atlas Permian), the contract is structured as a per-truck monthly fee plus a per-load variable. Specific pricing is not public. Revenue in 2024 was de minimis; the company guided to meaningful revenue only in 2026 from the Atlas expansion and new oil-and-gas pilots.
Traction over time
| Date | Milestone | Capital Raised (cum.) |
|---|---|---|
| 2018 | Founded | <$5M seed |
| 2019 | Series A $40M | $40M |
| 2021 | Series B $125M | $165M |
| 2022 | Series C $300M | $465M |
| 2024 | Debt + convertible notes | ~$500M |
| Apr 2025 | SPAC deal announced ($2.5B EV) | - |
| Aug 2025 | $100M pre-merger funding | ~$600M |
| Sep 2025 | Public via Ares Acquisition Corp II; +$275M | ~$875M |
| Oct 2025 | 8 driverless trucks live at Atlas Permian | - |
(Kodiak/press/SEC filings, 2018-2025)
Market analysis
The US for-hire trucking market is roughly $940B in annual revenue (ATA, 2024); long-haul freight is ~$430B of that, and off-highway haul (mining, oil/gas, forestry) adds another $25-40B. Driver shortage is the structural force: ATA estimated the US was short 60,000-80,000 drivers in 2024 and the gap could reach 160,000 by 2030. Driverless trucks don't reduce the cost of manned operations meaningfully — they eliminate the driver line item (~35-40% of total operating cost) and remove the hours-of-service limit, enabling 22-hour daily utilization vs. the 11 a human can legally run. Autonomous trucking's TAM depends on the multiple: at 10-20% share capture by 2035, revenues sit in the $40-90B range; at 5%, it's a specialty market for a handful of operators.
Competitive intel
Aurora Innovation (NASDAQ: AUR) — The clear category leader. $21B+ peak market cap; went commercial in April 2024 on its Dallas-Houston route, now operates ~10 driverless trucks with no safety driver on ~500 commercial loads/month. Deeper OEM integration (Paccar, Volvo) than Kodiak. Aurora has higher capital reserves (>$1.5B as of Q3 2025) and a bigger hardware footprint.
Waabi — Toronto-based; Raquel Urtasun (ex-Uber ATG chief scientist). $300M+ raised; Khosla, Uber, Nvidia. End-to-end neural-net architecture ("Waabi Driver") vs. Kodiak's more modular approach. Just announced driverless pilots with Volvo and Uber Freight in Texas (Waabi, 2025).
Plus (TuSimple fork) — Public via SPAC in 2024; rapidly pivoting to Driver-Assist L4+ for OEMs rather than pure L4.
Torc Robotics — Owned by Daimler Truck since 2019. Advantage of embedded OEM integration; disadvantage of single-OEM lock-in.
Gatik — Middle-mile B2B (grocery, retail); different route profile but overlaps for short-haul freight.
Chinese entrants (Pony.ai Freight, TuSimple China, Inceptio) — Not competing in US market but important for global TAM.
Internal programs at fleets — J.B. Hunt, Schneider, and Werner are all evaluating multiple vendors, which is why Kodiak's pilots exist. Could flip to in-house or single-vendor exclusivity.
History and evolution
2018: Founded by Don Burnette and Paz Eshel in Mountain View. 2019: Series A $40M. 2021: Series B $125M. 2022: Series C $300M led by SoftBank. 2023-24: Capital raising harder; shifts to debt and convertibles; announces oil-and-gas pivot with Atlas. April 2025: Announces de-SPAC merger with Ares Acquisition Corp II at $2.5B EV. August 2025: Secures $100M pre-merger (FleetOwner, Aug 2025). September 2025: Lists on NASDAQ as Kodiak AI (KODA). October 2025: Confirms 8 driverless trucks operational for Atlas in the Permian Basin, 3,000+ new highway commercial miles added in six months (FreightWaves, Oct 2025). The stumbles: lost its Waymo-Uber-era co-founder Paz Eshel to Battery Ventures in 2021; cash-raising conditions in 2023 forced down-rounds that didn't make headlines but weakened cap table.
What people say
The case for. FreightWaves' editorial consistently rates Kodiak as the only sub-Aurora player with real commercial traction (FreightWaves, 2024-2025). Atlas's willingness to run 8 trucks with no safety driver is the hard signal — before Kodiak, no commercial customer had agreed to zero-observer ops. Oil-and-gas reviews are positive on the economics: ATEX Safety and some Permian fleet managers have told trade press the Kodiak sand-haul lane shows 30%+ cost savings vs. conventional manned operations (FleetOwner, Aug 2025). Don Burnette is consistently cited as one of the more technical-depth CEOs in the AV category.
The complaints. Kodiak lags Aurora on on-highway commercialization — Aurora removed safety drivers from Dallas-Houston in April 2024, while Kodiak is targeting H2 2026 for the equivalent. Sell-side notes from BofA and Goldman (Sep 2025 post-SPAC) flag the cash runway as the primary risk, with ~$275M raised against burn that sell-side pegs at $150-180M/year. Short-interest on KODA climbed into double-digit percent of float within a month of listing (FactSet, Oct 2025). Plus, the Permian sand-haul market is narrow — critics argue it's a single-customer, single-geography deployment and doesn't prove the broader thesis.
Outlook: the open question
Does Kodiak hit meaningful driverless on-highway revenue in H2 2026 before its ~$275M post-SPAC cash reserve runs out? The SPAC capitalization pencils for roughly 18-24 months of runway at current burn, which lines up tightly with the H2 2026 driverless on-highway target. If Kodiak ships zero-observer highway operations on a commercial route (Dallas-San Antonio, Atlanta-Nashville) by year-end 2026, the stock rerates sharply and strategic/PIPE financing becomes easy. If the target slips another 12 months — which has happened repeatedly across AV — Kodiak has to raise at a discount, likely tapping Atlas or J.B. Hunt as strategic lifelines, which dilutes meaningfully. The key secondary signal is the highway miles-with-no-safety-driver count; Aurora reports this quarterly and Kodiak will have to as a newly public company. If Kodiak's quarterly count is <10% of Aurora's by Q2 2026, the thesis is losing.
How to attack it
The most credible wedge is vertical-specific off-highway autonomy — pick one of Kodiak's target verticals (forestry, metal mining, agricultural haul) and build a truck-plus-software-plus-operations company from scratch. Kodiak's modularity is a strength and a weakness: because its stack has to generalize across site types, it sacrifices the 10-20% performance gains a vertical specialist can engineer. Ex-Freeport-McMoRan or ex-Vale operations leaders paired with ex-Zoox/Cruise engineers could build an "autonomous haul truck for copper mines" company and sell to the top 20 miners globally. Second wedge: remote operations as a service — every AV trucking company runs a remote-ops center, and these are expensive, understaffed, and fragile. A staffed-remote-ops company that serves Kodiak, Aurora, Waabi, Nuro, Zipline, and Gatik would arbitrage the fragmentation (Phantom Auto already attempted this and was acquired; the thesis still holds).
Biggest exploitable weaknesses: (1) single deployment at Atlas — a Chevron Permian sand-logistics change or Atlas financial difficulty re-prices Kodiak's whole story; (2) modular stack is slower to specialize than Aurora's OEM-integrated approach, so per-mile margins may always trail; (3) Roush manufacturing dependence creates a single-supplier bottleneck for truck retrofits; (4) cash runway tight (~18-24 months); (5) public-market sentiment on AV is cyclical and 2027 could be a drought for secondary raises; (6) Waabi's end-to-end neural-net architecture could scale faster if 2026 autonomy benchmarks tilt neural-net; (7) regulatory — Texas and Arizona are permissive, but any state-level incident during driverless highway ops could freeze permits broadly.
Adjacent-segment play
The clear adjacent segment is autonomy-in-a-box for OEM retrofits — a product that lets any fleet operator (not just Kodiak customers) retrofit a Freightliner, Peterbilt, or Kenworth with an L4-capable stack. Kodiak partially does this but keeps the stack closed; an open-stack competitor (think: Open Source Tesla-Autopilot clone for trucks) could sell to the mid-tier fleet market that will never contract for Driver-as-a-Service. Second adjacency: commercial vehicle insurance — AV trucks need bespoke insurance, and the three or four reinsurers underwriting AV today are undercapitalized for the market. A Lemonade-for-autonomous-fleet-insurance company that uses Kodiak's and Aurora's telemetry would own a defensible book. Third adjacency: off-highway L4 for closed-site industrial applications — ports, airports, warehouses — all are closed-site, high-volume, lower-regulation deployments. A less-attractive adjacent: robotaxis (Kodiak has explicitly avoided this). Waymo, Cruise, Zoox, Pony own the robotaxi stack and the regulatory playbook, and trucking vendors who pivot usually underperform. Gatik already occupies the middle-mile B2B adjacent segment and is the clearest analog to Kodiak's off-highway thesis.
Sources and further reading
- FreightWaves, "Kodiak AI, now a public company, looks to deliver an autonomous trucking future", October 8, 2025 — https://www.freightwaves.com/news/kodiak-ai-now-a-public-company-looks-to-deliver-an-autonomous-trucking-future
- FleetOwner, "Kodiak Robotics secures nearly $100M in new funding ahead of SPAC merger, public debut", August 2025 — https://www.fleetowner.com/news/news/55318347/kodiak-robotics-secures-nearly-100m-in-new-funding-ahead-of-spac-merger-public-debut
- TechCrunch, "Autonomous trucking startup Kodiak Robotics to go public via SPAC", April 2025 — https://techcrunch.com/2025/04/14/autonomous-trucking-startup-kodiak-robotics-to-go-public-via-spac
- Robot Report, "Kodiak Robotics autonomous trucking developer goes public via SPAC", 2025 — https://www.therobotreport.com/kodiak-robotics-autonomous-trucking-developer-goes-public-via-spac/
- FreightWaves, "The leaders, followers and also-rans in autonomous trucking", 2024 — https://www.freightwaves.com/news/the-leaders-followers-and-also-rans-in-autonomous-trucking
- PitchBook Analyst Note, "Autonomous Trucking Gains Traction", 2024 — https://files.pitchbook.com/website/files/pdf/PitchBook_Analyst_Note_Autonomous_Trucking_Gains_Traction.pdf
- FreightWaves, "Can Kodiak Robotics compete with the big boys of autonomous trucking?", 2024 — https://www.freightwaves.com/news/can-kodiak-robotics-compete-with-the-big-boys-of-autonomous-trucking
- ATA, 2024 American Trucking Trends — https://www.trucking.org/economics-and-industry-data
Capital history
| Date | Round | Amount | Valuation | Lead(s) |
|---|---|---|---|---|
| 2018 | <$5M | — | — | |
| 2019 | $40M | — | — | |
| 2021 | $125M | — | — | |
| 2022 | $300M | — | — | |
| 2025-04 | $2.5B EV | — | — | |
| 2025-09 | ~$275M (incl. $145M PIPE + $62.9M trust) | — | — |
Investors / owners: Ares Management, Battery Ventures, SoftBank Vision Fund, CRV, Lightspeed, ARK Invest, George Soros, Cathie Wood
Competitive set
- Aurora Innovation (NASDAQ: AUR) — Category leader; driverless commercial on Dallas-Houston since April 2024
- Waabi — Toronto; Raquel Urtasun; end-to-end neural-net approach
- Plus — Public via SPAC 2024; pivoting to L4+ driver-assist
- Torc Robotics — Owned by Daimler Truck; OEM-captive
- Gatik — Middle-mile B2B; different route profile
- Pony.ai (Pony AI Freight) — China-US dual listing; cross-border AV