KBR deal gives UK AI start-up Applied Computing a North Sea route to market
Shell contractor's partnership with Applied Computing opens a distribution channel that sidesteps the operator-by-operator sales cycle dogging energy AI adoption.
US engineering firm KBR has partnered with UK start-up Applied Computing, announced on Monday (2026-07-20), placing the Aberdeen-based company's machine-learning verification tools inside KBR's North Sea service portfolio.4 The agreement gives Applied Computing its first major oil and gas industry validation and a route to operators — including Shell — through KBR's existing framework relationships.
The energy industry has been slow to deploy AI on live production assets. Operators worry that a model trained on one reservoir will fail without warning when geology shifts, or that a decommissioning-cost algorithm will drift from its training data without anyone noticing. Applied Computing founder Harsh Tukra has put it plainly: "Trust is not built as a feature, it's earned. One recommendation, one shift at a time."4 That framing resonates on a continent shelf where proof-of-concept projects routinely stall before reaching the control room.
KBR's endorsement changes the commercial logic for Applied Computing. Winning individual North Sea operators requires separate procurement cycles, legal review and field-by-field integration work; KBR's framework agreements compress that process significantly.4 The start-up now has a deployment path through a contractor already embedded in Shell operations and other major North Sea producers.
The timing places Applied Computing inside a market undergoing active portfolio reshuffling. Ithaca Energy completed the purchase of a 50 percent stake in two Shell licenses in the West of Shetland Basin on Friday (2026-05-22), while simultaneously farming down its Fotla discovery.1 Shell remains a major operator in that basin, and KBR's service contracts in the area will now incorporate Applied Computing's verification layer.
The trust problem the start-up addresses is practical, not theoretical. Oil and gas AI models must generalise across fields with different rock properties, fluid compositions and production histories. A corrosion-risk model calibrated on a mature gas platform may produce unreliable outputs on a high-pressure, high-temperature well without recalibration.4 Applied Computing's approach lets engineers audit model confidence for each specific prediction, rather than relying on a single lab-derived accuracy figure.
Shell's own position adds context. The company stated on Wednesday (2026-06-03) that it had around $20 billion of capital employed across lower-carbon platforms, including power, hydrogen and carbon capture.2 Late-life asset decisions — when to sell, when to extend, when to decommission — are increasingly AI-assisted, and the reliability of those models directly affects the economics of those calls.
KBR's decision to partner rather than acquire suggests the technology sits at a stage where it is credible enough to offer clients but not yet ready to fold into a proprietary stack.4 For Applied Computing, the arrangement provides distribution without the due diligence exposure of a full sale; for KBR, it adds a differentiated capability without development cost.
Separately, North Sea decommissioning activity has been drawing new entrants. Boskalis Subsea Services signed a preferred partnership frame agreement with Anasuria Operating Company on Tuesday (2026-07-08), covering inspection, repair and maintenance, decommissioning and construction work.3 That deal illustrates the contractor competition intensifying around late-life North Sea infrastructure — the same market KBR is positioning to serve with Applied Computing's tools.
The adoption risk is real. A partnership announcement creates distribution access; it does not guarantee that Shell or other operators will accept the verification layer as standard practice across their fields. Tukra's own framing — "one recommendation, one shift at a time" — implies a slow build rather than a step-change in deployment.4 The indicator worth tracking is whether Applied Computing's tools move from framework agreement to live field deployment within a specific named contract in the next two quarters. Until that happens, the KBR deal is a commercial credential, not yet a revenue line.