A hub-and-spoke network that delivers density and speed without owning $52B in trucks and terminals.

Every major LTL carrier in the United States runs the same model: own the tractors, own the trailers, own 200 to 600 terminals per network, hire 15,000 to 39,000 employees, and route freight through hub-and-spoke. The model is so universal that "hub-and-spoke" and "LTL carrier" are treated as synonyms.

They are not synonyms. The shape is right. The ownership is what broke.

This piece walks through why hub-and-spoke is mathematically the right network shape for palletized freight, why fleet ownership stopped being the moat it used to be, what changed between 2018 and 2026 that made an unowned network possible, and how a four-layer model delivers the same density and speed without the $700M-per-year capex bill that defines incumbent LTL.

1. Why Hub-and-Spoke Was the Right Shape

Hub-and-spoke beats point-to-point at scale. The math is well understood. If you have N origin-destination pairs and route every shipment direct, you need N dedicated lanes. If you route through a single hub, you need 2N lanes (one origin-to-hub, one hub-to-destination) but each lane can be aggregated across many shippers.

For LTL freight, where individual shipments rarely fill a 53-ft trailer, aggregation is the entire game. A pallet from Phoenix to Atlanta does not justify a dedicated truck. Combine it with 22 other pallets going east through Dallas, and the per-pallet cost drops by an order of magnitude.

This is why every major LTL carrier converged on the same shape. FedEx Freight runs 355 terminals. Old Dominion runs ~260. XPO runs 614 locations. Saia runs 214. TForce runs ~658. ArcBest runs ~240. The shape is not the problem. If you are running palletized freight at scale, you will end up with hubs.

2. Why Owning the Fleet Became the Wrong Constraint

The original logic for fleet ownership was capital scarcity. In the 1980s and 1990s, when most major LTL carriers built their networks, the things you needed to run a national carrier (trailers, tractors, real estate, dock workers, dispatch software, EDI) were either expensive, scarce, or both. Owning them was the moat.

By 2026, every component on that list is either commoditized, available on demand, or both:

• Tractors and trailers: A liquid spot market with 250,000+ active power units and persistent oversupply. Asset utilization, not asset access, is the constraint.
• Real estate: 50+ shared cross-dock facilities can be operated as third-party space, with capacity scaled per shipment instead of per square foot of owned terminal.
• Labor: Cross-dock labor is a contracted service in every major metro. The dock worker scarcity that defined the 2010s loosened materially in 2023-2026.
• Software: Dispatch, tracking, billing, and accessorial-classification software exists as APIs and SaaS. The internal-build moat collapsed.

What ownership now produces is the opposite of leverage: a fixed-cost base that crushes operating ratios when volume softens, and a capacity floor that cannot flex below the owned-fleet utilization threshold. Old Dominion spent $771 million in capex in 2024 and watched its operating ratio deteriorate 410 basis points in two quarters. The capex was not the cause of the deterioration; it was the cause of why the deterioration mattered. A network with no owned fleet would have flexed down with the volume.

3. What Changed Between 2018 and 2026

Five things made the unowned-network model possible at scale, none of which existed at sufficient maturity before 2018:

1. API-native carriers. By 2024, every major brokerage and asset-light carrier exposed dispatch, status, and rating endpoints. A routing engine can now book capacity in real time across thousands of carriers without phone calls or EDI translation layers.
2. Mobile driver tooling. Driver apps with GPS, photo POD, and real-time exception reporting eliminated the visibility gap that used to require carrier-owned telematics. A managed network now has the same shipment-level visibility as an asset-based carrier.
3. Cross-dock as a service. Shared dock facilities with neutral operators emerged in every major metro. A network can route through 50+ cross-docks without owning any of them, at variable cost per pallet instead of fixed cost per square foot.
4. Capacity glut. The 2022-2026 freight recession created persistent oversupply in tractor and box-truck markets. A network can buy capacity at marginal rates that asset-based carriers cannot match without bleeding their own utilization.
5. Decision-quality software. The routing problem (mode, lane, carrier, dock, schedule) is now solved by software in seconds, with quality that exceeds dispatcher decisions for most shipment classes. Owning the dispatcher used to be the moat. The dispatcher is now a model.

Each of these alone would not have been enough. Together, they removed the structural reasons fleet ownership produced advantage.

4. The Four-Layer Network Model

An unowned hub-and-spoke network has four layers. Each layer can be operated independently, but the leverage compounds when all four are integrated.

Layer 1: Shipper Interface

The shipper-facing booking, quoting, tracking, and exception management surface. This is what shippers experience as "the carrier." It includes self-serve quote tools, an API and CLI, an MCP server for AI agents, account-level integrations, and a control tower for visibility. The shipper does not see (or need to see) which carrier hauled the linehaul or which cross-dock the freight passed through.

Layer 2: Decision Layer

The routing engine that decides per shipment: which mode (LTL, box truck, cargo van, FTL, pool), which lane, which middle-mile carrier, which cross-dock, which schedule. This is the layer that replaces the dispatcher and the pricing department in a traditional carrier. Decisions are made in seconds and updated continuously based on real-time capacity and rate signals.

Layer 3: Middle-Mile Capacity

The pool of carriers that move freight between cross-docks and to final delivery. 30 LTL carriers and 20,000+ FTL, box truck, and cargo van capacity sources, contracted at network rates and integrated into the decision layer via API. The network does not own any of this capacity, but contracts and routing logic make it behave as one capacity pool.

Layer 4: Cross-Dock Network

50+ cross-dock facilities across 18 markets. These are the hubs. They serve the same role as carrier terminals (consolidation, deconsolidation, pool distribution, cross-dock transfer), but they are operated as shared facilities, not owned terminals. Capacity scales with demand. The freight does not sit waiting for a terminal-utilization window before it moves.

5. The Density Math: How a Network Builds Density Without Owning Trucks

The argument against unowned networks has always been density: how can you aggregate enough volume to make hub-and-spoke economics work if you do not control the trucks?

The answer is that density is a routing problem, not an ownership problem. When the decision layer can see all available capacity (owned or contracted) and all available freight (across all shippers in the network), it can match them at the per-shipment level instead of the per-truck level.

Two structural advantages compound:

• Cross-shipper aggregation. An asset-based carrier can only aggregate freight from shippers who chose that carrier. A network can aggregate freight from every shipper in the platform, regardless of which carrier ultimately moves it. Density per lane is higher because the funnel is wider.
• Multi-modal substitution. An LTL carrier can only route freight through LTL. A network can route the same shipment through pool distribution, box truck, cargo van, or FTL depending on origin density, destination density, and equipment fit. 78% of Warp shipments use equipment other than 53-ft trailers, which is not possible in a single-mode owned network.

The result is denser per-lane utilization than the asset-based incumbents achieve, despite having no owned tractors. 1,400+ LTL lanes have crossed the density threshold where direct dispatch beats hub-routing, which is the same threshold ODFL and FedEx Freight optimize for in their own networks.

6. The Touch Math: 1-2 vs 4-5 Handoffs

Damage and delay in LTL are functions of how many times freight is handled. A traditional hub-and-spoke shipment touches 4 to 5 facilities: origin pickup, origin terminal, line-haul, hub terminal, destination terminal, destination delivery. Each touch is a damage opportunity and a dwell window.

An unowned cross-dock network reduces this to 1 to 2 touches per shipment by routing freight directly between cross-docks when density supports it, and by using shared dock facilities that consolidate at the cross-dock instead of moving through terminal hubs. The freight is in motion more, sitting less, and handled fewer times.

Result on 655,767 completed shipments: 0.81% damage rate vs the 1.24% industry baseline (ATA claims data), a 31% reduction. 98.2% on-time delivery vs an industry average that no major incumbent publishes but is widely understood to sit between 92 and 96%.

7. What This Model Does Not Replace

The unowned-network model is not the right answer for every freight type. Three categories where asset-based carriers retain structural advantage:

• Hazmat at scale. Hazmat certifications, specialized equipment, and compliance regimes favor carriers that own the asset and the certification. Warp ships hazmat through its FTL network only; LTL hazmat consolidation is not the right fit for an unowned model.
• Reefer at sustained volume. Refrigerated freight requires temperature-controlled cross-docks, sealed-trailer chain-of-custody, and specialized driver training. These are best aligned with carriers that build and operate the entire reefer stack.
• Single-shipper dedicated capacity. When a shipper has consistent enough volume to fill a dedicated fleet, the network premium does not apply. Dedicated capacity belongs in dedicated contracts with asset-based providers.

The model replaces traditional LTL for the majority of palletized commercial freight. It does not replace specialized modes that depend on asset-specific capabilities.

8. Industry Parallels: Stripe, Airbnb, AWS

Three other industries went through the same transition from asset-ownership-as-moat to network-as-moat:

• Payments. Banks owned the ledgers and the processing infrastructure. Stripe built a network layer on top of existing banking rails and now processes more payment volume than any individual bank.
• Hospitality. Hotel chains owned the buildings, the staff, and the brand. Airbnb built a network on existing residential supply and now exceeds Marriott in nightly inventory.
• Compute. Companies owned servers, racks, and data centers. AWS built a network on existing power, fiber, and silicon supply and now operates more compute than any individual enterprise.

In each case, the underlying physical assets did not disappear. The asset owners did not all go out of business. What changed was where the leverage moved: from "who owns the assets" to "who orchestrates the network." The same dynamic is now playing out in freight.

9. What This Means for Shippers Right Now

If you are evaluating LTL alternatives in 2026, the practical implications are concrete:

• Per-pallet pricing replaces tariff lookups. A network model prices per pallet with all-inclusive rates because it does not need to defend a fixed terminal cost base. No fuel surcharges, no accessorials, no reweigh-reclass disputes.
• Mode flexibility on the same booking. The same network handles LTL, pool distribution, box truck, cargo van, and FTL through one booking surface. You stop running RFPs across five carrier categories.
• API and agent access without an account manager. Quote, book, and track via API, CLI, or MCP server. The integration is a 3-minute key, not a 90-day implementation.
• Cost reduction without RFP cycles. The network optimizes per shipment in real time, so cost reductions accrue continuously instead of at annual contract negotiation.

The structural argument is straightforward: the hub-and-spoke shape is right, and an unowned network delivers it without the cost base that makes the incumbents' rates inflexible.

Frequently Asked Questions

Is hub-and-spoke without a fleet actually a different network model, or just brokerage?

It is a different model. Brokerage matches a single shipment to a single carrier on a single lane. A hub-and-spoke network without a fleet routes shipments through shared cross-docks, aggregates freight across shippers at the lane level, and orchestrates the entire flow through software. The end result is a network that behaves like an asset-based LTL carrier from the shipper's perspective, but does not carry the fixed-cost base. Traditional brokerage does not include the cross-dock layer or cross-shipper aggregation.

How can a network achieve LTL density without owning the trucks?

Density comes from how freight is aggregated, not from who owns the equipment. When a network can pool freight across all shippers in the platform and route through shared cross-docks instead of carrier-specific terminals, the funnel feeding each lane is wider than any single asset-based carrier can achieve on the same lane. Combine this with multi-modal substitution (using box truck or pool distribution where they fit better than LTL) and the network achieves higher per-lane utilization than the incumbents on most lanes.

Does this model work for enterprise shippers or only mid-market?

Both. Mid-market shippers benefit most immediately because they get enterprise-grade visibility, pricing, and service without enterprise-grade contract minimums. Enterprise shippers benefit from the model on lanes where their asset-based carriers struggle with capacity, density, or accessorials, particularly long-haul, multi-stop, or non-loading-dock destinations. The model also serves enterprise needs for API integration, agent access, and real-time visibility that legacy LTL carriers cannot provide at the same maturity.

What happens when freight volume spikes? Does an unowned network have surge capacity?

Surge capacity is one of the structural advantages of the model. An asset-based carrier is capped at the utilization of its owned fleet. A network can flex capacity by activating additional carriers from the contracted pool, with rates that adjust per lane as supply tightens. During peak season and weather events, the network model has historically shown more surge capacity than asset-based carriers, not less, because the supply pool is larger than any single carrier's owned fleet.

How does the cost compare to traditional LTL?

Comparison across 568+ lanes against the published tariffs of the same incumbents (FedEx Freight, Old Dominion, XPO, Saia, TForce, ArcBest) shows 24% lower all-inclusive cost on average. The savings come from three sources: no fixed terminal cost base to amortize, no fuel surcharges (rates are quoted all-inclusive), and elimination of accessorial and reclass-reweigh fees that account for 8 to 15% of total LTL spend in traditional contracts.

Does this model handle long-haul, or only regional?

Both. The 1,400+ LTL lanes covered include cross-country routes (LA to NYC, Seattle to Atlanta, etc.) as well as regional and intra-metro. Long-haul lanes are routed through the cross-dock network with consolidation at origin and deconsolidation at destination, which is mechanically similar to how asset-based carriers route long-haul through their hubs. The difference is the cross-docks are shared facilities operated at variable cost per pallet rather than owned terminals operated at fixed cost per square foot.

What about damage and claims? Is the network responsible if a contracted carrier damages freight?

Yes. The network is the contracting party with the shipper and is responsible for damage claims regardless of which carrier hauled the freight. This is the same responsibility model as a traditional LTL carrier. Damage rates on the network (0.81% across 655K+ shipments) are 31% lower than the 1.24% industry baseline because the touch count per shipment is lower, not because claim handling is different.

Why have the incumbent LTL carriers not adopted this model?

Two reasons. First, the asset base is the constraint they are built around. A carrier with $3 to 7 billion in owned equipment and 200 to 600 owned terminals cannot transition to an unowned model without writing down the asset base, which their balance sheets and shareholders will not absorb. Second, the operating model is built around terminal utilization. Routing freight outside owned terminals reduces utilization on the existing network, which is the metric every operating ratio depends on. The incumbents are structurally locked into the model that defines them.

Network performance data based on Warp operating data: 655,767 completed shipments, 50+ cross-dock facilities across 18 markets, 30 LTL carriers, 20,000+ FTL/box truck/cargo van capacity. Industry damage baseline (1.24%) sourced from American Trucking Associations claims data. Cost comparisons reference published LTL tariffs from FedEx Freight, Old Dominion, XPO, Saia, TForce, and ArcBest. Capex and operating ratio figures sourced from 2024 carrier annual reports and earnings calls.