Rail unions oppose a historic merger. Here’s how AI logistics can improve rail safety, transparency, and service reliability—without disruptive consolidation.
Rail Merger Pushback: AI Can Fix Rail Without M&A
A proposed $85 billion Union Pacific–Norfolk Southern merger is being pitched as a simple fix: create the first true coast-to-coast rail network, remove interchanges, and ship faster. Two of the largest U.S. rail unions aren’t buying it. Their argument is blunt: consolidation can raise safety risk, weaken service for smaller markets, and leave shippers with fewer real options.
I think the unions are pointing at something the industry rarely says out loud: network efficiency isn’t the same as network consolidation. You can reduce dwell, smooth handoffs, and improve on-time performance without stitching two giants together. The practical way to do that in 2026 isn’t another mega-deal—it’s AI-driven rail operations and logistics optimization that makes constraints visible, decisions auditable, and plans resilient.
This post is part of our “AI in Supply Chain & Procurement” series, where we focus on how AI improves planning, sourcing, risk, and execution. Rail is a perfect stress test because it sits at the intersection of physical infrastructure, labor rules, safety, and customer commitments—exactly the kind of multi-variable environment where good AI (and good governance) pays off.
Why unions oppose the transcontinental rail merger
Answer first: The unions opposing the UP–NS merger are signaling that the biggest risks aren’t theoretical—they’re operational: safety, staffing, and service quality during and after integration.
The Brotherhood of Locomotive Engineers and Trainmen (BLET) and the Brotherhood of Maintenance of Way Employes Division (BMWED) oppose the deal and frame it as a “debt-ridden tie-up” that could make rail less competitive with trucking. They’re also warning about longer trains, slower operations, and the likelihood of shedding or transferring lines that serve small towns, farms, and factories to short lines. In plain terms: mainline efficiency could come at the expense of network coverage and local service reliability.
From a supply chain and procurement perspective, this matters because rail service is often embedded in long-term contracts for bulk commodities, chemicals, agricultural products, and intermodal moves. When network changes create unpredictable transit times or fewer routing options, procurement teams end up paying in three places:
- Higher freight rates (less competition, more pricing power)
- Higher inventory buffers (to protect service variability)
- Higher expediting costs (when rail misses the plan and truck becomes the backstop)
The railroads’ counterpoint is also straightforward: an end-to-end network reduces interchange delays in bottlenecks like Chicago. That’s a real operational issue. But it’s not a merger-only issue.
The merger promise vs. the real bottleneck: coordination
Answer first: Interchange friction is mainly a coordination problem—data, timing, yard capacity, and accountability—not a physics problem that only a merger can solve.
Mergers can reduce handoffs because the handoff happens “inside” one company. But shippers and rail customers don’t actually care who owns the rails; they care whether the move is predictable.
The painful part of interchange is usually some combination of:
- Trains arriving when a yard is already saturated
- Crews and power not aligning with arrival windows
- Work queues not prioritized to customer commitments
- Limited visibility into ETA confidence (not just ETA)
Here’s the stance I’ll take: if a merger is your main tool to fix coordination, you’re admitting your operating system is the constraint. That’s exactly where AI-based logistics systems can help—especially when they’re integrated with planning, procurement commitments, and real-time execution.
What “AI coordination” looks like in rail terms
AI doesn’t need to “run trains.” It needs to run decisions—the thousands of micro-choices that create dwell, missed connections, and cascading delays.
A practical AI stack for rail coordination typically includes:
- Predictive ETAs with confidence bands (e.g., “arrival 08:40–09:25, 80% confidence”) rather than a single timestamp.
- Constraint-aware scheduling that accounts for yard capacity, crew availability, maintenance windows, and track time.
- Dynamic prioritization tied to customer SLAs, demurrage exposure, and downstream plant/warehouse impacts.
- Exception prediction (what will break before it breaks) to prevent missed interchanges.
When that system works, interchange becomes less of a gamble. And when interchange becomes predictable, the merger rationale weakens.
Safety and labor concerns: AI works only if it’s transparent
Answer first: AI can reduce safety risk and ease labor concerns, but only when it’s built around transparency, auditability, and worker-facing tooling—not management-only dashboards.
The unions’ safety concerns aren’t just about incident rates. They’re about how decisions get made under pressure—train length policies, maintenance deferrals, staffing ratios, and the real-world incentives that push risk downhill.
AI can help, but the design has to be intentional. Here’s what I’ve found works in operations environments with strong labor rules: make the model explain itself in operational language.
AI-powered monitoring that supports safety (and proves it)
A credible approach combines multiple data streams:
- Track geometry and wayside detector feeds
- Work order and inspection histories
- Speed restrictions, weather, and territory risk profiles
- Crew schedules and fatigue risk signals (policy-compliant)
Then the AI produces outputs like:
- “This segment’s failure risk is elevated because… ” with contributing factors
- Maintenance prioritization lists that show tradeoffs explicitly
- Near-miss and anomaly detection with a clear threshold logic
The point isn’t to replace safety programs. It’s to make safety decisions consistent and reviewable.
Job guarantees aren’t a plan—operating plans are
UP has cited “jobs-for-life” commitments with some unions. Labor groups opposing the deal argue guarantees don’t cover realities like seniority protections and relocation costs.
AI can’t negotiate contracts. But AI can reduce fear by giving everyone access to the same operational truth:
- If a network change is proposed, the model can estimate the impact on crew starts, terminal staffing, and maintenance-of-way workload by geography.
- Scenarios can be reviewed and challenged: “Show me what happens to staffing if you increase average train length by X%,” or “If you reroute traffic off this line, what happens to transit time and local service?”
That’s a healthier conversation than “trust us after the merger.”
Shippers’ worry: fewer choices and higher rates
Answer first: Shippers opposing rail consolidation are reacting to a simple procurement reality—when you lose competitive tension, you usually lose pricing discipline.
Agriculture and chemicals groups have raised concerns about reduced competition and disruption. A competing railroad has also voiced competitive concerns. This is predictable: rail is already a constrained market in many lanes.
For procurement leaders, the bigger issue is that rail service quality and rail pricing are tightly linked. When service becomes unreliable, procurement teams either:
- Pay premiums for guaranteed capacity
- Shift volume to truck (often at a cost)
- Add inventory (working capital hit)
AI can protect shippers without changing ownership
If you’re a shipper, you can’t control rail M&A. But you can demand (and build) a data-driven operating relationship. Specifically:
- Service scorecards based on appointment adherence, dwell by node, and variance—not averages
- Lane-level forecasts that translate your demand plan into carload/intermodal needs 2–8 weeks out
- Shared exception workflows so disruptions trigger concrete actions (replans, resequencing, alternative terminals)
This is where the “AI in supply chain & procurement” theme shows up clearly: better forecasting and better supplier performance management reduce your exposure to consolidation risk.
A better path than consolidation: AI-enabled “virtual single line” rail
Answer first: The strongest alternative to a mega-merger is a “virtual single line” experience—multiple carriers operating like one through shared data, aligned incentives, and AI-assisted planning.
Railroads already do interline moves. The difference between a painful interline move and a smooth one is whether everyone has:
- the same plan,
- the same constraints,
- and the same definition of “late.”
AI makes that alignment feasible at scale by turning fragmented operational data into shared, decision-ready visibility.
What to implement (even if you’re not a railroad)
If you’re a shipper, 3PL, or manufacturer trying to stabilize rail performance in 2026, these are the highest ROI moves I’d prioritize:
- Predictive arrival + inventory impact
- Tie ETA confidence to production schedules, plant run rates, and warehouse labor plans.
- Order-to-rail translation
- Convert purchase orders and demand plans into carload needs and cutoff dates early enough to matter.
- Disruption playbooks that the AI can trigger
- Pre-approved alternates: transload, alternate ramps, mode shift thresholds, customer communication templates.
- Network “digital twin” scenarios
- Run “what-if” on reroutes, terminal closures, and peak surges without waiting for real-world failure.
These aren’t science projects. They’re operating discipline, supported by AI.
People also ask: practical questions procurement teams should be asking
Will AI reduce the need for rail mergers?
Often, yes—because it reduces the coordination penalty that mergers claim to solve. If interchange, yards, and capacity planning become predictable, “single-line” becomes less valuable.
Does AI create labor risk?
It can, if it’s used as a blunt instrument for headcount cuts. Used properly, AI creates workload clarity and safer planning—and it gives unions and frontline workers a way to challenge assumptions with data.
Where do you start if you don’t control the rail network?
Start with what you own: demand signals, appointment discipline, exception management, and supplier performance scorecards. Then require visibility and variance reporting from partners.
Next steps: make rail efficiency measurable before it’s political
Merger debates turn into narratives fast—jobs, safety, prices, “competitiveness with trucks.” The fastest way to improve outcomes for everyone (workers, shippers, consumers) is to turn rail efficiency into measurable, shared metrics and use AI to close the gap.
If you’re leading supply chain or procurement, don’t wait for the Surface Transportation Board process to dictate your risk posture. Build a plan that assumes volatility and rewards predictability: better forecasts, better rail supplier management, and better exception response.
Where would you rather invest: a decade of integration risk from consolidation, or a year of AI-enabled operating improvements that make service—and accountability—visible?