EV Battery JV Breakups: What AI Sees Before You Do

Ford and SK On didn’t “break up” because one side suddenly stopped believing in EVs. They broke up because the math changed.

On Dec. 18, 2025, the companies confirmed they’re dissolving their BlueOval SK battery joint venture. The original plan—announced in 2021—was a massive $11.4B bet on three U.S. battery plants: two in Kentucky and one in Tennessee. Under the dissolution agreement, Ford will own the Kentucky plants and SK On will own and operate the Tennessee facility.

If you work in supply chain or procurement, don’t file this away as “auto industry drama.” This is a clean case study in what happens when demand signals, policy incentives, and capital plans drift out of sync—and why AI-driven supply chain risk management is becoming table stakes for any company making decade-long commitments.

The BlueOval SK dissolution: what actually changed

The straightforward answer: the capacity Ford expected to need in 2026+ no longer matches the EV market reality it’s seeing in 2025.

The joint venture was built for scale: three large battery plants intended to feed Ford’s future EV lineup. But several hard constraints stacked up:

• EV demand cooled relative to the aggressive forecasts baked into 2021–2022 investment decisions.
• The U.S. federal EV tax credit (up to $7,500) ended on Sept. 30—a major demand lever disappearing overnight.
• Ford has been delaying EV launches while expanding hybrids, pushing planned battery production timelines (Kentucky now eyed for 2026).
• Ford’s EV unit has been burning cash: Model e lost $5.1B in 2024 after $4.7B in 2023, and it’s still losing money in 2025.

Meanwhile, SK On is signaling a sharper focus on “profitable and sustainable growth” and more diversified demand: supplying other customers and even stationary energy storage from the Tennessee plant.

This matters because joint ventures are supposed to reduce risk. In reality, they can also concentrate risk when both partners are locked into one shared set of assumptions.

A procurement lens: JVs fail quietly long before the announcement

The press release is the last step. The real dissolution started earlier, when leading indicators diverged:

• Forecast error widened (EV demand vs. plan)
• Utilization scenarios fell below break-even
• Policy and incentive assumptions weakened
• Cash cost per kWh didn’t fall as fast as modeled

If your category plan depends on a “forever partnership,” you’re already exposed.

Why EV battery supply chains punish slow decision-making

The blunt answer: batteries combine commodity volatility, industrial-scale capex, and policy-driven demand. That’s a nasty mix.

Even for sophisticated teams, battery supply is harder than most direct materials because:

• Inputs are globally interdependent (lithium, nickel, manganese, graphite, separators, electrolyte)
• Qualification cycles are long (cell chemistry, pack design, thermal systems, safety testing)
• Capacity is lumpy (you don’t add 5% capacity; you build a plant)
• Policy can flip demand fast (incentives, domestic content rules, fuel economy targets)

BlueOval SK also sat in the middle of government financing: a $9.6B Department of Energy loan closed in Dec. 2024 to support construction of the three plants.

Once you’re in that zone—big loans, multi-year construction, workforce ramp, and model launch dependencies—your supply chain isn’t “flexible” unless you designed it that way on purpose.

The myth procurement needs to drop

Myth: “If we pick the right partner, supply is stable.”

Reality: Stability comes from optionality—multiple demand scenarios, multiple customer paths, renegotiation triggers, and operational playbooks that don’t assume linear growth.

That’s where AI helps—not as a buzzword, but as a practical way to manage complexity at the speed the market changes.

How AI could have flagged the risk earlier (and what it should monitor)

The direct answer: AI doesn’t predict a JV breakup like a fortune-teller. It detects the conditions that make a breakup rational.

In the “AI in Supply Chain & Procurement” series, I keep coming back to one theme: AI is best when it watches weak signals continuously, then forces humans to deal with them before they become headlines.

Here are the signal clusters an AI-driven risk and planning stack should track for partnerships like BlueOval SK:

1) Demand sensing beyond your own order book

Traditional S&OP often overweights internal forecasts. For EVs, that’s dangerous.

AI demand sensing can blend:

• Dealer inventory and days-on-lot (where available internally)
• Price elasticity signals (transaction prices vs. MSRP, incentives)
• Macro indicators (interest rates, consumer sentiment)
• Competitive launches and pricing pressure

When these signals move against your baseline forecast, AI can quantify how much capacity becomes stranded and when.

2) Policy and incentive monitoring as a first-class input

A lot of companies treat policy shifts like “news.” They should be treated like demand drivers.

In this case, the end of the EV tax credit on Sept. 30 changes affordability calculations immediately. AI systems can:

• Model demand impact by vehicle segment
• Stress-test volume plans by region
• Trigger contingency scenarios (hybrid mix shift, phased ramp)

For procurement, the key is connecting policy changes to contracted volumes, take-or-pay exposure, and capex ramp commitments.

3) Supplier and partner financial/operational posture

SK On explicitly said it wants to realign assets for efficiency and supply other customers. That’s not negative—it’s rational. But for Ford, it’s a signal: your supply plan depends on a partner whose optimal path is diversifying.

AI can help by continuously scoring:

• Capacity allocation risk (who else needs the cells?)
• Timeline risk (construction, commissioning, yield ramp)
• Contract flexibility risk (renegotiation probability)

4) “Break clauses,” renegotiation triggers, and negotiation timing

Most companies store contracts like PDFs and only re-open them when there’s a problem.

Contract AI (applied responsibly) can:

• Extract key terms (volume commitments, ownership, change control)
• Map obligations to operational plans
• Alert teams when forecast variance approaches renegotiation thresholds

The practical win: you renegotiate from a position of preparation, not panic.

Snippet-worthy truth: If your contract only gets read when the relationship is failing, you don’t have contract management—you have document storage.

What procurement leaders should do next (a practical playbook)

The answer: build optionality into supplier strategy and use AI to keep it honest.

You don’t need to be building battery plants to learn from this. Any category with big capex, long lead times, or regulatory exposure—semiconductors, chemicals, aerospace, medical devices—has similar dynamics.

Step 1: Create a “partnership stress test” template

Run this quarterly for major suppliers and JVs:

• Demand downside scenarios: -10%, -25%, -40%
• Policy shock scenarios (incentive removal, tariff changes, new compliance rules)
• Utilization break-even points by facility
• Minimum cash burn tolerance before governance changes

If you can’t quantify the downside, you’re not managing it.

Step 2: Segment suppliers by switching friction, not spend

Spend-based segmentation misses what matters.

Add friction variables:

• Requalification time (months)
• Tooling and engineering dependency
• IP entanglement
• Single-site vs. multi-site risk

High-friction suppliers require earlier triggers and more redundancy.

Step 3: Build a multi-customer capacity clause mindset

SK On wants to supply other customers and stationary storage. That’s a common evolution.

Procurement should negotiate for:

• Transparent allocation rules during shortages
• Capacity reservation options
• Pricing tied to clear indices (with caps/collars)
• Defined governance for major demand mix shifts

Step 4: Use AI for early warnings, not annual “risk workshops”

Most companies get this wrong: they schedule risk reviews like annual physicals.

Instead:

• Monitor leading indicators weekly
• Tie alerts to specific actions (renegotiate, dual-source, inventory hedge)
• Assign an owner per risk with a deadline

A warning without an owner is just noise.

What this signals for 2026 planning (and why it’s timely right now)

The direct answer: 2026 is shaping up to be the year of capacity realism.

Across the auto industry, the mood has shifted from “build as fast as possible” to “build what we can profitably use.” You can see similar moves elsewhere, including other automakers scaling battery production and adjusting timelines.

For supply chain leaders heading into year-end planning (and the budget resets that come with January), BlueOval SK is a reminder that:

• Your 3–5 year plan needs scenario ranges, not a single line
• Supplier partnerships need governance that anticipates change
• AI in procurement is most valuable when it reduces decision latency

If you’re still relying on quarterly spreadsheets to spot structural shifts, you’ll find out about your risk when the market does.

A better way to approach supplier partnerships in volatile markets

Ford and SK On say they’ll remain strategic partners and continue battery supply—even as they unwind the JV ownership structure. That’s the point: dissolving the JV doesn’t mean the relationship disappears. It means the structure is being adjusted to fit new constraints.

In the “AI in Supply Chain & Procurement” series, this is the pattern to pay attention to: supplier relationships are becoming more dynamic, and procurement teams are being asked to manage change without breaking service, compliance, or cost targets.

If you want one practical next step, it’s this: stand up an AI-assisted supplier risk dashboard for your top 20 relationships—the ones where switching is hardest and failure is most expensive. Then connect that dashboard to contract terms and S&OP scenarios, so alerts translate into real decisions.

What would your biggest partnership look like if demand drops 30% next quarter—or if an incentive disappears at the end of the month? If you can answer that crisply, you’re already ahead.