Why AI matters at this scale

HF Sinclair is a large, independent petroleum refiner and marketer operating refineries and logistics networks across the United States. The company's core business involves converting crude oil into essential products like gasoline, diesel, and jet fuel, which it markets under brands like Sinclair. With a workforce of 5,001–10,000 employees, HF Sinclair manages capital-intensive, continuous-process manufacturing assets where operational efficiency, safety, and margin optimization are paramount.

For an enterprise of this size in the oil and energy sector, AI is a critical lever for maintaining competitiveness and navigating energy transition pressures. The scale of operations generates vast amounts of real-time sensor, process, and transactional data. Leveraging this data with AI can drive step-change improvements in predictive maintenance, supply chain agility, and energy efficiency, directly impacting the bottom line. At this employee band, the company has the financial resources and operational complexity to justify significant AI investment, but must navigate the challenges of integrating new technology into legacy industrial environments and upskilling a large workforce.

Concrete AI Opportunities with ROI Framing

1. Predictive Maintenance for Critical Assets: Refineries run 24/7, and unplanned downtime can cost over $1 million per day. AI models analyzing vibration, temperature, and pressure data from rotating equipment can predict failures weeks in advance. This allows maintenance to be scheduled during planned turnarounds, avoiding catastrophic failures. The ROI is direct: reduced maintenance costs, extended asset life, and increased production availability, with project paybacks often under two years.

2. Crude & Supply Chain Optimization: Refinery margins are heavily influenced by crude selection and logistics. AI can continuously analyze global crude assays, real-time freight costs, and product market prices to recommend the optimal crude slate and logistics plan. This dynamic optimization can improve gross margin by cents per barrel, which, multiplied by hundreds of thousands of barrels per day, translates to tens of millions in annual EBITDA uplift.

3. Process & Energy Efficiency: Distillation and cracking units are massive energy consumers. Machine learning can identify non-intuitive optimal setpoints for process variables (temperatures, pressures, flow rates) by learning from historical operational data. This can reduce energy consumption by 2-5%, significantly cutting costs and Scope 1 & 2 emissions. The ROI comes from lower natural gas and power bills, alongside potential credits for emissions reduction.

Deployment Risks Specific to This Size Band

For a company with 5,001–10,000 employees, AI deployment faces specific scale-related risks. Organizational inertia can be high, with decision-making slowed by multiple layers of management and the need to align disparate business units (refining, logistics, retail). Data fragmentation is acute, as legacy systems (OSIsoft PI, distributed control systems) and IT/OT silos exist across large, geographically dispersed sites, making unified data lakes challenging. Change management becomes a massive undertaking, requiring training thousands of engineers and operators on new AI-driven workflows. Finally, cybersecurity risk escalates as connecting operational technology to AI platforms expands the attack surface of critical national infrastructure, demanding robust, and often costly, security frameworks.