How oil and gas companies deploy AI
Oil and gas companies deploy AI mostly in production, drilling and exploration, and count the returns in barrels: BP's Optimization Genie adds about 2,000 barrels a day at its Atlantis hub, and Occidental Petroleum's AI adds 2,500 barrels of oil equivalent a day. TotalEnergies' CEO said he was skeptical about AI in the subsurface, and months later TotalEnergies committed more than €100 million to a lab with Mistral that builds AI models for exploration. AI often sits far from IT: ConocoPhillips' CFO oversees digital technology, and Chevron put enterprise AI under the president of New Energies.
Verified See methodology
How oil and gas companies have deployed AI
- Scaling
Saudi Aramco
Integrated oil, gas and chemicals
aramcoMETABRAIN is the industrial large language model Saudi Aramco built and trained on decades of company data, launched in Mar 2024; Saudi Aramco said in Nov 2025 that it answers about 40,000 prompts a day.
$2.6B
AI-driven value (2025)
- Piloting
BP
Integrated oil and gas
BP builds its own AI tools, such as Optimization Genie, the Wells Assistant and PlantPulse, has used Palantir Foundry since 2014 and added Palantir's AIP in a 5-year deal in Sep 2024, and set out in Aug 2025 to merge about 100,000 data pipelines into one platform with Databricks and Palantir by mid-2026, a platform BP was still building in Dec 2025 and has not reported complete.
~2,000 b/d
Optimization Genie production gain
- Piloting
Chevron
Integrated oil and gas
Chevron builds its own AI tools: ApEX, a multi-agent generative AI system for exploration data launched in Aug 2024, APOLO for shale production forecasts, and ScoutAI for basin assessments (Oct 2026).
$100M+
AI value captured in 2026
- Piloting
Occidental Petroleum
Oil and gas exploration and production
Occidental builds its AI applications in-house through the Applied AI Center of Excellence, led by Vice President and Chief of AI Patrick Bangert (Sep 2026) and set up by Feb 2025, with focused teams for operations, the Gulf of America, and logistics and supply chain (Aug 2025).
2,500 boe/d
Annual production uplift from AI plunger lift optimization
- Experimenting
ConocoPhillips
Oil and gas exploration and production
ConocoPhillips has named no enterprise AI platform: its AI runs as models inside operational and commercial workflows, built by product teams that pair business experts with Chief Digital & Information Officer Pragati Mathur's staff, and by a data science and AI group led by Bo Hu.
Every well, daily
Gas-lift wells AI checks
- Experimenting
ExxonMobil
Integrated oil, gas and chemicals
ExxonMobil builds proprietary AI workflows in-house on its own well and seismic data, and is moving the whole company onto one SAP-based process and data platform, with larger rollouts planned for 2027 (Jul 2026).
50,000
Industry wells screened by AI for exploration
- Experimenting
Shell
Integrated oil, gas and energy
Shell runs its industrial AI on the Shell.ai Platform, built on C3 AI on Microsoft Azure since Sep 2018 with Databricks for data and extended under a multi-year C3 AI deal in Jun 2026; Shell also built Shell e, a platform on which teams build generative AI applications.
13,000+
Equipment under AI predictive maintenance
- Experimenting
TotalEnergies
Integrated energy: oil, gas, LNG and electricity
TotalEnergies builds its own AI applications in the Digital Factory, about 500 digital and AI experts in France and India who have built nearly 100 solutions since 2020, on industrial data platforms bought from Cognite and AspenTech (2025 annual report, Mar 2026).
10,000+
Live AI models in operation
Findings
What the oil and gas profiles show
Every return an oil company credits to a single AI tool comes from production optimization: BP's Optimization Genie added about 2,000 barrels a day at the Atlantis hub in the Gulf of America (first published Apr 2025, restated Feb 2026), and Occidental Petroleum credits AI plunger lift optimization with 2,500 barrels of oil equivalent a day and AI gas lift optimization with a 3% annual production uplift (Sep 2026). ConocoPhillips said in Mar 2026 that AI gas-lift checks cut site visits, without a figure.
Saudi Aramco and Chevron are the only oil companies with a firm-level figure credited to AI, and neither publishes a return per use case: Saudi Aramco's 2025 Annual Report (Mar 2026) puts $2.6 billion of its $5.3 billion Technology Realized Value on AI-driven solutions, an internal metric of gains over a business baseline verified by an unnamed third party, and Chevron CFO Eimear Bonner said on Oct 30, 2026 that AI had captured more than $100 million of value so far in 2026, without saying how Chevron measures it.
The sector's large cost programs give AI no share: ExxonMobil credits about 75% of its $16.9 billion of structural cost savings since 2019 to organizational initiatives (Oct 2026), none of Shell's $5.1 billion of structural cost cuts from 2022 to 2025 or the $700 million more in the first half of 2026 is credited to AI, BP has not said how much of its $4 billion of structural cost reductions by Oct 2026 came from AI, and neither TotalEnergies' $7.5 billion savings program for 2026 to 2030 nor ConocoPhillips' more than $1 billion cost program is credited to AI.
Oil companies mostly build their own AI tools: Saudi Aramco trained aramcoMETABRAIN on decades of company data (launched Mar 2024), Chevron built ApEX for exploration data, APOLO for shale forecasts and ScoutAI for basin assessments (Oct 2026), BP built Optimization Genie, the Wells Assistant and PlantPulse on Palantir and Databricks, Occidental Petroleum builds through its Applied AI Center of Excellence (Sep 2026), and TotalEnergies' Digital Factory of about 500 digital and AI experts has built nearly 100 solutions since 2020 (Mar 2026). Shell is the exception, running its industrial AI on the Shell.ai Platform, built on C3 AI since Sep 2018 and extended in Jun 2026.
Saudi Aramco is the only oil company to report company-wide employee use of AI: Ahmad Al-Khowaiter said on Aug 31, 2026 that tens of thousands of Saudi Aramco employees use its in-house language models. BP counts single tools instead, more than 2,000 users with access to its wells and subsurface assistants (Jul 2025) and 400 users of its Databricks maintenance agents (Jun 2026); the 30,000 Microsoft 365 Copilot licenses at TotalEnergies come from a Microsoft customer story (Oct 2025), not from TotalEnergies; and Chevron, ExxonMobil, Shell, ConocoPhillips and Occidental Petroleum have not said how many employees use AI tools.
Stages
Where oil and gas companies stand
Scaling
Scaling means the company reports returns it credits to AI in several functions, or a firm-level AI figure, and runs a company-wide AI platform with a named owner.
Piloting
BP, Chevron, Occidental Petroleum
Piloting means the company reports a business return it credits to AI in at least one use case, or a company-wide AI rollout with a user count and a named owner.
Experimenting
ConocoPhillips, ExxonMobil, Shell, TotalEnergies
Experimenting means the company has reported neither a business return it credits to AI nor a company-wide AI rollout with a user count. Pilots, partnerships, plans and spending alone keep a company at Experimenting.
Returns
The returns oil and gas companies credit to AI
Only realized returns the company reported and credited to AI itself, with their dates: revenue, cost, headcount, losses avoided or output per employee.
| Company | Use case | Return |
|---|---|---|
| BP | Optimization GenieUpstream | About 2,000 barrels a day more production at the Atlantis hub (Feb 2026)Source |
| Occidental Petroleum | Plunger lift optimizationU.S. Onshore operations | 2,500 boe/d annual production uplift, credited to AI plunger optimization (Sep 2026) |
| Occidental Petroleum | Gas lift optimization networkU.S. Onshore operations | 3% annual production uplift from AI gas lift optimization (Sep 2026) |
Compare
Oil and gas side by side
| Company | Ownership | Platform | Models | Hosting | Reach | Stage |
|---|---|---|---|---|---|---|
| Saudi Aramco | EVP Technology & Innovation Ahmad Al-Khowaiter, on the Executive Management team | Hybrid: METABRAIN built in-house; DeepSeek, OpenAI and partner platforms alongside | In-house METABRAIN models, plus DeepSeek and OpenAI models | Own Dhahran data centers (Dammam-7, Nvidia SuperPODs); Groq inference center in Dammam via Aramco Digital; Microsoft Azure under study | Tens of thousands of employees on in-house models (Aug 2026) | Scaling |
| BP | Emeka Emembolu, EVP Technology, on the leadership team | Hybrid: In-house tools; Palantir Foundry and AIP, Databricks and Copilot bought | Microsoft Copilot and Palantir AIP; models behind most in-house tools not named | Microsoft Azure (upstream data) and AWS (retail data), Databricks across both, Palantir Foundry, and BP's own Houston high-performance computing center on NVIDIA GPUs | Not disclosed company-wide; many teams use generative AI daily (Oct 2025) | Piloting |
| Chevron | Jeff Gustavson, President of New Energies; Brent Gros from Jan 1, 2027 | Hybrid: ApEX, APOLO and ScoutAI built in-house; Copilot, Claude, SLB and DataRobot tools bought | Microsoft Copilot and Anthropic Claude in use; models behind in-house tools not disclosed | Microsoft Azure, Chevron's primary cloud, including seismic imaging on Azure HPC; DataRobot inspection agents run at the edge on Chevron sites; ApEX hosting not disclosed | Not disclosed; AI used across the business (Jan 2026) | Piloting |
| Occidental Petroleum | VP and Chief of AI Patrick Bangert and an AI Center of Excellence; the CIO reports to the CEO | Hybrid: AI applications built in-house; AWS cloud and SensorUp emissions platform bought | In-house machine learning, computer vision and agents; no model provider named | AWS, Occidental's preferred cloud since Jul 2023; where the AI models run is not disclosed | AI runs 7,100 onshore wells (Sep 2026); employee use not disclosed | Piloting |
| ConocoPhillips | CFO and SVP Digital and Transformation Konnie Haynes-Welsh (since Sep 2026) | Hybrid: No named AI platform; own models plus vendor tools | No provider named; models for drilling, gas lift and forecasting | Not disclosed; filings cite unnamed third-party cloud providers, and reservoir modeling was set to move to SLB's cloud-based DELFI in 2022 | Not disclosed; drilling, production, reservoir and commercial teams | Experimenting |
| ExxonMobil | SVP Neil Chapman, whose technology remit includes AI; IT sits inside the central technology organization | Hybrid: Proprietary AI built in-house; SAP data platform and Azure cloud bought | Proprietary in-house models; no outside LLM provider disclosed | Microsoft Azure for Permian field data and machine learning (since 2019); other AI hosting undisclosed (the Discovery 6 seismic supercomputer is not described as AI) | Not disclosed; no employee AI tool or user count published | Experimenting |
| Shell | Global Head of AI Amy Challen; CIO Robbert van Rutten reports to the CEO | Hybrid: Shell.ai Platform on bought C3 AI and Databricks; Shell e built for generative AI | In-house ML models; GPT-4 via Azure OpenAI; Microsoft Copilot; a NeMo-tuned chemistry LLM | Microsoft Azure for C3 AI and Azure OpenAI; Databricks on AWS and Azure; SLB subsurface tools on AWS; Shell's own HPC cluster in Bangalore; NVIDIA edge devices at sites | No employee AI user count; latest figure an AI community of 12,000+ members (Jan 2025) | Experimenting |
| TotalEnergies | Namita Shah, President OneTech, on the Executive Committee; CDO Frédéric Gimenez runs the Digital Line | Hybrid: Digital Factory builds in-house; Copilot, Cognite and AspenTech bought | In-house ML models, Microsoft 365 Copilot, Mistral frontier models in development | Microsoft Azure and AWS clouds (partners since 2021), in-house supercomputers in Pau; Mistral model hosting not disclosed | 30,000 Copilot licenses (Microsoft, Oct 2025) | Experimenting |
Questions
AI in oil and gas, answered
- Which oil and gas companies are furthest along with AI?
- In the Enterprise AI diffusion report, Saudi Aramco is at Scaling, BP, Chevron, and Occidental Petroleum are at Piloting, and ConocoPhillips, ExxonMobil, Shell, and TotalEnergies are at Experimenting. A stage follows from the returns and rollouts a company has publicly reported in the last 18 months, not from its spending or plans.
- How do oil companies use AI?
- Oil companies use AI most in production, drilling, exploration and equipment maintenance. Occidental Petroleum runs about 7,100 U.S. Onshore artificial lift wells with AI (Sep 2026), ExxonMobil uses AI to direct Permian drilling paths (Dec 2025) and drilled a Guyana well autonomously with AI (May 2026), Chevron reports a 50% cut in DJ Basin lost-production events (Oct 2026), and Shell's C3 AI predictive maintenance covers more than 13,000 pieces of equipment (Jun 2026). Saudi Aramco also runs its own industrial language model, aramcoMETABRAIN, which answered about 40,000 prompts a day by Saudi Aramco's count in Nov 2025.
- What return have oil and gas companies reported from AI?
- Oil and gas companies have reported few returns they credit to AI. Saudi Aramco credits AI-driven solutions with $2.6 billion of its 2025 Technology Realized Value (Mar 2026), Chevron credits AI with more than $100 million of value so far in 2026 (Oct 2026), BP's Optimization Genie added about 2,000 barrels a day at the Atlantis hub (first published Apr 2025, restated Feb 2026), and Occidental Petroleum credits AI plunger lift optimization with 2,500 barrels of oil equivalent a day (Sep 2026). ExxonMobil, Shell, ConocoPhillips and TotalEnergies have reported no return they credit to AI since May 2025: ExxonMobil's Dan Ammann said in Sep 2025 that AI is helping the bottom line without a figure, and TotalEnergies' Patrick Pouyanné described 1% to 2% more plant availability as an aim in Feb 2026, not a result.
- Which AI models do oil and gas companies use?
- Oil and gas companies pair in-house models with a few outside providers. Saudi Aramco runs its own METABRAIN models alongside DeepSeek and OpenAI models, Chevron staff use Microsoft Copilot and Anthropic's Claude (Jun 2026), BP uses Microsoft Copilot and Palantir AIP, Shell uses GPT-4 through Azure OpenAI and Microsoft Copilot, and TotalEnergies uses Microsoft 365 Copilot and is developing models with Mistral for its subsurface program (Sep 2026). ExxonMobil builds proprietary models and has disclosed no outside model provider, and Occidental Petroleum and ConocoPhillips have named no model provider.