JBS USA × Völur — AI carcass-sorting and cutting-plan optimisation at Hyrum UT (Proof-of-Value phase)

NA-US (Hyrum UT initial deployment; planned scale-out to other JBS USA plants)

Content

JBS USA, the US arm of JBS S.A. (the world’s largest meat processor), announced in January 2023 a partnership with Völur to deploy Völur’s AI for carcass-sorting and cutting-plan optimisation. The initial Proof-of-Value ran for 6 months at JBS USA’s Hyrum, Utah beef plant, with the Völur model hosted on Microsoft Azure.

What the AI does:

• Optimises daily carcass sorting and cutting plans
• Balances supply (carcass quality / grade mix) against demand (customer orders) in real time
• Maximises carcass value (matches parts to highest-value orders)
• AI is on the cutting-plan stage (downstream of the slaughter floor), not on the carcass-quality grading itself

Deployment status as of mid-2026:

• Hyrum POV: completed
• Plans announced to scale to other JBS USA plants at end of POV
• Post-2023 status not publicly confirmed — no trade press confirmation surfaced that Völur is in scaled production at additional JBS USA plants
• Treat as pilot-stage with scale ambition until verified

Why this is structurally different from Cargill CarVe and Tyson × AWS:

• Cargill CarVe = computer vision on carcass yield (line-side, real time)
• Tyson × AWS = computer vision on poultry-piece counting (line-side, real time)
• JBS × Völur = decision-support AI for cutting-plan optimisation (not on the line, in the planning layer)
• Together these three cover most of the surface area of NA meat-processing AI

What this unit is doing in the taxonomy

Anchors the processing × predictive ML × input-reduction cell for NA meat processing — the planning-layer counterpart to Cargill CarVe and Tyson × AWS (line-layer). Sister units:

• Cargill CarVe (cargill-carve-meat-processing.md) — beef, line-side CV
• Tyson × AWS (tyson-aws-poultry-vision.md) — poultry, line-side CV
• SoraLINK (soralink-export-food-processing.md) — Canadian NA processing AI, predictive maintenance

Why it matters for talks

• JBS completes the NA meat-packing trifecta. Cargill, Tyson, and JBS are the three largest US protein packers. All three now have publicly documented AI deployments — but in three different cells: line vision (Cargill), line vision (Tyson), cutting-plan optimisation (JBS). Worth distinguishing in talks.
• The “planning layer” AI is a different shape from “line layer” AI. CarVe and Tyson × AWS affect yield in real time at the cut. Völur affects value capture across the day’s orders. The two have different feedback-loop speeds and different unit economics.
• Microsoft Azure is the AI compute substrate for Völur at JBS. Same hyperscaler pattern as Tyson × AWS (Amazon) and Cargill CarVe (Cargill in-house; some cloud partnerships).
• POV state is the honesty check on the knowledge base. The subagent’s recon was careful to note that scaling status is “pilot-stage with scale ambition, not verified at scale.”

Critical context

• All deployment evidence is from Völur and JBS press releases; the Hyrum POV completion is documented, but the subsequent scale-out is not.
• “Optimises” / “maximises” vendor framing — independent verification of carcass-value capture uplift not surfaced.
• Völur is Norway-origin with US business; the company is small relative to JBS USA. Single-vendor dependency is a strategic consideration worth naming.
• Data flows: cutting-plan AI requires detailed daily carcass / order data. Whether Völur retains aggregated cross-plant data is not publicly characterised.