Logistics · Deep dive

C.H. Robinson

The 120-year-old produce broker that became America's largest freight middleman — and is now cutting a quarter of its people while gaining share, betting that AI agents make the broker stronger, not obsolete.

well positioned

The disintermediation thesis has been tried and has failed twice — and C.H. Robinson is the rare incumbent that has turned the technology meant to kill it into a 22% headcount reduction, flat gross margins and twelve consecutive quarters of share gains through the worst freight recession in decades.

Snapshot

C.H. Robinson is the largest freight broker in the United States and has been for so long that much of the industry’s vocabulary comes from it. It owns almost no trucks. It sits between roughly 83,000 shipper customers and a contracted network the company puts at about 450,000 carriers (company site, 2026), matching loads to trucks and keeping the difference. FY2025 total revenues were $16.2B, down 8.4% — a near-meaningless number. The one that matters is adjusted gross profit: $2.7B in 2025, what is left after paying the carrier. On that base it ran a 29.7% adjusted operating margin excluding restructuring in Q1 2026, having removed roughly 22% of its workforce in two years. Market cap is about $18.3B (MarketBeat, mid-2026). The company is now a single wager: that AI agents let one broker do the work of two, and that the broker is still needed at all.

Founding story

Charles Henry Robinson was a New Yorker and a travelling salesman out of St. Louis who moved his family to Grand Forks, North Dakota after the railroad reached it, on the theory that settlers on the northern plains would want fruit they could not grow. He began brokering produce into the Red River Valley in the early 1900s, and in May 1905 folded the business into a partnership with Nash Brothers, the state’s leading wholesaler and the ancestor of Nash Finch. The entity was incorporated as C.H. Robinson Co.

The original problem was perishables: get produce from grower to grocer before it spoils, using rail and wagons the company did not own. That constraint — solve the logistics without owning the assets — became the genetic code. Robinson Fresh, still a segment today ($37.5M of adjusted gross profit in Q1 2026), descends directly from that produce desk.

The company spent most of the 20th century as an employee-owned, branch-office affair, then renamed itself C.H. Robinson Worldwide and listed on NASDAQ in 1997, raising roughly $190M — for the ~101 employee shareholders who sold, not for the company. It has never needed outside capital since. The ownership story after that is simple: institutions bought it, and it bought itself back.

How it works

A shipper needs 40,000 lb moved from Ohio to a DC in Texas by Thursday. It does not want a fleet, and its carrier relationships are thin outside core lanes.

Robinson takes the order — increasingly by email, parsed by an LLM straight into the order system — prices it, and finds a truck. About 70% of NAST truckload volume in Q1 2026 moved under contract on pre-negotiated rates and a committed routing guide; 30% moved spot, versus 65/35 a year earlier. When the primary carrier rejects a load it falls to the next name in the guide. Average routing-guide depth in Managed Solutions was 1.5 in Q1 2026 against 1.3 a year earlier — an unglamorous way of saying capacity is tightening.

Finding the truck means reaching a network of mostly tiny carriers and getting one to accept a rate. Historically that was a human on a phone, calling until someone said yes. Now it is increasingly a fleet of agents inside Navisphere: quote, order, truck-post, load-booking, tracking, document and carrier-payment agents (Q1 2026 earnings deck).

The economics are a spread. In Q1 2026 North American truckload price per mile rose 11.0% Y/Y while cost per mile rose 13.0% — capacity tightened faster than Robinson could reprice — and it still held NAST gross margin flat at 14.6%, on dynamic pricing and what it calls a widening cost-of-hire advantage. That flat line, in a quarter where costs outran prices by 200bps, is the most important number the company produced this year.

Product and business overview

North American Surface Transportation (NAST). Truckload, LTL, intermodal and drayage. Q1 2026 adjusted gross profit of $431.1M, up 3.0%, on $2.95B of revenue. LTL is quietly the star — gross profit per order up 8.5% Y/Y on 2.0% volume growth, while truckload volume fell 3.5%. Roughly two-thirds of company gross profit, and effectively the entire equity story.

Global Forwarding. Ocean, air and customs, built on the 2012 Phoenix International deal. Q1 2026 adjusted gross profit of $162.3M, down 12.1%, as ocean rates collapsed against rising vessel capacity (ocean volume -10.5%, air tonnage -15.0%). Customs was the bright spot, up 20% on tariff complexity. Ancora wanted this segment sold; management kept it and defends its margin instead (24.4%, up 60bps).

All Other and Corporate. Robinson Fresh (the 1905 business) and Managed Solutions, a TMS-plus-services offer with $1.9B of freight under management in Q1 2026 and adjusted gross profit up 6.3%. Europe Surface Transportation was sold on 1 February 2025 and contributes zero.

Navisphere. The platform: shipper visibility and quoting, a free carrier booking-and-tracking app, and the substrate the AI agents run on.

Business model and pricing

There is no price list. Robinson sells a spread, and the spread is the product.

Book it correctly and the arithmetic is stark. Q1 2026 total revenues were $4.01B; $3.02B went straight out to carriers as purchased transportation, plus $337M of purchased produce. Adjusted gross profit — revenue minus what third parties are paid — was $660.5M, a 16.5% gross margin. Income from operations was $175.7M. Against revenue that is a 4.4% operating margin, which is why gross-revenue coverage makes brokers look like doomed low-margin businesses. Against adjusted gross profit — the real revenue line of an agent business — it is 26.6%, or 29.7% excluding restructuring, and 37.4% in NAST.

The only real lever on that margin is labour, so the metric management lives by is shipments per person per day; NAST president Michael Castagnetto said in October 2025 it was up about 40% versus 2022. Q1 2026 average headcount: 4,752 in NAST (-10.0% Y/Y), 3,848 in Global Forwarding (-14.8%), 3,105 in corporate (-12.6%), with $18.8M of severance booked in the quarter. The stated model is to decouple headcount growth from volume growth entirely — a polite way of saying volume can double and hiring will not.

For carriers, the one published price point is QuickPay: payment in about three days for a 1.5% fee, against standard terms measured in weeks.

Traction over time

The through-line: revenue has fallen by a third since the 2022 peak, and profitability per unit of gross profit has risen every year anyway. Transportation adjusted gross margin ran 13.8% in 2021 and 17.5% in 2025. Headcount fell from roughly 14,990 in Q1 2024 to roughly 12,085 by Q4 2025 (FreightWaves and Trucking Dive, on company disclosures). NAST volume outpaced market indices for the twelfth consecutive quarter in Q1 2026, against a 6.2% decline in the Cass Freight Shipment Index. On 29 October 2025 the company raised its 2026 operating income target to $965M-$1.04B. Capital returns are accelerating: $360M in Q1 2026, more than double the prior year, plus a dividend raised for 28 consecutive years (MarketBeat, 2026).

Market analysis

Estimates of the US freight brokerage market cluster around $19-20B in 2025 on a net-revenue basis: Mordor Intelligence puts it at $19.5B growing to $29.1B by 2030 (8.4% CAGR); Market.us says $19.9B. On that framing Robinson’s ~$2.7B of adjusted gross profit implies low-to-mid-teens share of a deeply fragmented market where thousands hold broker licences and the long tail is one person with a phone.

Three forces move at once. The freight recession is in its fourth or fifth year, with the Cass shipment index still down 6.2% Y/Y in Q1 2026. Capacity is finally tightening — Robinson’s truckload cost per mile rose 13% — helped by an FMCSA crackdown on chameleon carriers, CDL mills and ELD fraud that is squeezing out the marginal capacity suppressing rates. And most consequentially, the cost of the broker’s own labour is collapsing: voice and language agents now handle check calls, quoting and load booking. That deflates the industry’s cost base, and it is genuinely ambiguous whether the benefit accrues to the incumbent with the data or to an entrant with no legacy payroll.

Competitive intel

The named set is in the frontmatter; what matters is the shape of the attacks. TQL, the #3 broker at roughly $6.8B of 2024 gross revenue, competes with a large human sales floor — precisely the model Robinson is dismantling. Landstar’s agent network has almost no fixed labour to automate, making it both the safest and the most exposed. Uber Freight has 30-plus AI agents and infinite patience and, after a decade, is still one player among many.

Convoy is the most instructive competitor and it no longer exists. It raised $260M at a $3.8B valuation, promised to automate the broker away, and shut down on 19 October 2023 (CNBC); its assets went to Flexport for roughly $16M, the platform onward to DAT in July 2025. The disintermediation thesis has been funded, staffed and tested. It lost.

The unresolved threat is not a competitor but a supplier. HappyRobot — $44M Series B in September 2025 at a reported ~$500M valuation, with eight of the ten largest brokers reportedly as customers (Upstarts, 2025) — sells the AI labour layer to everyone. If a five-person brokerage can buy Robinson-grade agentic coverage off the shelf, the scale moat compresses to pricing data, carrier trust and balance sheet. Those are real. They are narrower than they were.

History and evolution

• 1905 — Robinson partners with Nash Brothers in Grand Forks, ND; C.H. Robinson Co. incorporated to broker produce.
• October 1997 — Renamed C.H. Robinson Worldwide; IPOs on NASDAQ, raising ~$190M for ~101 selling employee shareholders.
• 2012 — Buys Phoenix International for roughly $635M, creating Global Forwarding.
• January 2015 — Buys Freightquote.com from Great Hill Partners for $365M — the online LTL/SMB channel.
• 2017-2019 — Bolt-ons: Milgram (Canada), The Space Cargo Group (Spain/Colombia, ~EUR 42M).
• 2021-2022 — Freight boom; revenue peaks near $24.7B and headcount balloons.
• 2022 — Ancora presses the board to refocus on the domestic core and consider selling Global Forwarding, and wins board seats. CEO Bob Biesterfeld departs in December, widely read as an Ancora outcome.
• June 2023 — Dave Bozeman appointed CEO; the lean operating model is imported wholesale.
• October 2023 — Convoy shuts down, taking the pure digital-broker thesis with it.
• 2024 — Castagnetto takes NAST (January); Damon Lee joins as CFO from GE Aerospace (July); the December investor day sets the $350-450M incremental operating income target.
• February 2025 — Europe Surface Transportation divested.
• October 2025 — 2026 operating income target raised to $965M-$1.04B on the strength of Lean AI.
• 12 February 2026 — Algorhythm Holdings, a former karaoke company, claims its SemiCab platform lets one operator handle 2,000+ loads a year against a ~500 benchmark. CHRW falls 24% intraday, RXO -20%, Landstar -16% (Overdrive).
• April 2026 — Q1: adjusted EPS +15.4%, twelfth straight quarter of NAST share gains, $360M returned to shareholders.

What people say

The case for. Sell-side is broadly constructive: roughly a Buy consensus across ~24 analysts with an average target near $192 (TipRanks/Public.com, 2026), with Baird lifting its target to $230. Baird’s Dan Moore called the February 2026 selloff an unsubstantiated fear trade largely devoid of substance, arguing that useful AI needs proprietary data accumulated across cycles and incumbents are the ones who have it (Overdrive, 24 February 2026). Moore added the underrated point that load theft and fraud have made a known, bonded counterparty more valuable than ever — a moat that gets stronger as the industry digitises. Operationally, the case is the tape: gross margin flat in a quarter where carrier costs rose 200bps faster than prices, twelve straight quarters of share gains against a shrinking freight index, and management raising 2026 targets mid-recession. Carriers on TruckersReport consistently cite one thing in Robinson’s favour: they get paid, and QuickPay lands in about three days for 1.5% — cheaper than factoring.

The complaints. The employee reviews are the ugliest part of the story and have been consistent for decades. Robinson holds about 3.3 out of 5 on Glassdoor across ~3,857 reviews, with only ~51% recommending it to a friend (Glassdoor, 2026) — poor for a company of its stature. Recurring themes: whiteboarded KPIs and cold-call quotas (reviewers describe 60+ prospecting calls a month), a cliquey sales-floor culture, comp that does not track performance, and — post-Bozeman — a sharp rise in complaints about surveillance, RTO enforcement, short-term metric obsession and collapsing morale, several naming the CEO directly. Five straight years of layoffs will do that. Carriers are the second aggrieved constituency: threads on TruckersReport and expediter forums describe Robinson loads as the cheap-and-heavy ones nobody else wanted, rates compressed below breakeven on some lanes, and onboarding rejections driven by fraud paranoia. The structural bear case is the one the market voted on in February 2026: if agentic AI drives the cost of covering a load toward zero, the spread goes with it, and a business whose moat is 450,000 phone relationships becomes one whose moat is an API. DAT’s Ken Adamo put the sharpest version of it — AI may be a cheat code against ever needing scale.

Outlook: well positioned or at risk?

Well positioned — but for a narrower and more precarious reason than the bulls state.

The reason is not that AI can’t do brokerage. It plainly can do large parts of it — Robinson is proving that by removing a fifth of its workforce and shipping agents that book loads, chase tracking and pay carriers. The reason is that the deflation lands on the incumbent’s cost base first, and Robinson is the only broker at scale converting it into reported margin rather than a pitch deck. Adjusted operating margin ex-restructuring went from 27.6% to 29.7% in a year; NAST sits at 37.4% against a 40% target. It is gaining share, in a shrinking market, while shrinking itself.

The bear case has already been run as a controlled experiment: Convoy raised $260M at $3.8B, promised exactly this, and liquidated. The friction that kills automated brokerage is not algorithmic — it is fraud, detention, damaged freight, a driver who won’t answer the phone, and a shipper who needs someone liable at 2am. Robinson is bonded, capitalised and known, and in a market where hundreds of carriers have just been stiffed by collapsing brokers, that counts for more, not less.

What would change the call: if HappyRobot-class tooling genuinely commoditises coverage, the spread compresses industry-wide and Robinson’s 14.6% NAST gross margin — not its headcount — starts falling. That is the number to watch, and it has not moved. Two other risks are real: Global Forwarding is a melting ice cube in an ocean-rate collapse (AGP -12.1% in Q1 2026), and the company is extracting productivity from a workforce that is, by its own reviews, exhausted and afraid. You cannot run a churn-and-burn culture indefinitely on people you are simultaneously replacing.

But the position compounds — the data deepens, cost-to-serve falls, share gains stack. Twelve consecutive quarters is not a fluke. The middleman everyone keeps declaring dead is currently the most profitable it has ever been per dollar of gross profit.

Sources and further reading

• Q1 2026 Earnings Presentation (Form 8-K) — C.H. Robinson / SEC, 29 April 2026. Segment adjusted gross profit, headcount, margins, truckload price/cost per mile, AI agent roster.
• Will AI kill brokerages? Landstar, C.H. Robinson, RXO stocks tanked in ‘scare trade’ — Overdrive, 24 February 2026. The Algorhythm/SemiCab selloff, with Baird and DAT rebuttals.
• C.H. Robinson Raises 2026 Operating Income Target — C.H. Robinson, 29 October 2025. The $965M-$1.04B target and Lean AI bridge.
• C.H. Robinson Reports 2025 Fourth Quarter Results — C.H. Robinson, 28 January 2026. FY2025 revenue and adjusted gross profit.
• Headcount falls at C.H. Robinson as automation, AI reshape brokerage — FreightWaves, 2025. The multi-year headcount decline and productivity framing.
• C.H. Robinson Digs in Against Activist Investor Ancora — SupplyChainBrain, 2022. The activist campaign that preceded the CEO change.
• Bezos-backed freight firm Convoy shuts down — CNBC, 19 October 2023. The failed disintermediation experiment.
• HappyRobot Raises $44 Million To Bring AI Agents To Logistics — Upstarts Media, September 2025. The voice-AI layer selling into eight of the top ten brokers.
• C.H. Robinson Reviews — Glassdoor, accessed 2026. ~3.3/5 across ~3,857 reviews; the recurring culture themes.
• History of C.H. Robinson Worldwide, Inc. — FundingUniverse. The 1905 Grand Forks founding and Nash Brothers partnership.