Verified
Exelon
Experimenting- Utilities
- 20,571 employees (2025)
- SPARK, built in-house
- POSEIDON at all six utilities
- No AI owner named
- No AI ROI disclosed
Exelon deploys AI through an AI Center of Excellence, SPARK, an in-house platform on which employees develop and test AI ideas, and models its analytics teams build for its utilities, led by POSEIDON storm restoration estimates and an underground cable failure model. Baltimore Gas and Electric runs AI drone inspections built with Deloitte and NVIDIA. Exelon has published no AI spending, usage counts or AI-credited returns, and CEO Calvin Butler said in Sep 2026 he would not discuss internal AI plans.
exeloncorp.comTheir approach to AI
- The platform
- SPARK, an in-house AI platform, lets Exelon employees develop and test AI ideas, and Exelon says it built on its AI Center of Excellence in 2025 to scale AI in a more structured way (2025 Sustainability Report, Sep 2026).
- The models
- Exelon has not named the models behind SPARK or POSEIDON, and its analytics teams build their own outage and cable failure models; Baltimore Gas and Electric's drone inspections run vision models on NVIDIA Jetson modules through Deloitte's OptoAI platform, trained with NVIDIA Omniverse synthetic data.
- The reach
- POSEIDON runs at all six Exelon utilities (Dec 2025), but Exelon has not said how many of its 20,571 employees use SPARK or any other AI tool.
- The catch
- Exelon has published no AI spending, usage counts or AI-credited returns, and Calvin Butler said in Sep 2026 that he was not going to talk about how Exelon will deploy AI inside its business.
Snapshot
- Employees
- 20,571
- Dec 31, 2025; 8,656 covered by collective bargaining agreements (2025 Form 10-K)
- Technology budget
- Not disclosed
- Exelon reports a $41.7B capital plan through 2029 (Oct 2026) but no IT or AI budget
- AI/ML headcount
- Not disclosed
- An AI Center of Excellence and Exelon Analytics data science teams exist; sizes not published
- Who owns AI
- No named AI owner
- IT and cybersecurity sit under EVP and Chief Customer and Technology Officer Tim Peterson (since Feb 2026)
- Platform
- SPARK, built in-house
- Employees develop and test AI ideas on it (2025 Sustainability Report, Sep 2026)
- Models
- Not disclosed
- In-house outage and cable models; drone inspection uses NVIDIA-based vision models through Deloitte's OptoAI
- Reach
- Not disclosed
- POSEIDON live at all six utilities (Dec 2025); no SPARK user count
- Use cases
- Storm, grid, customer, safety, cyber
- Exelon names AI in storm response, cable failure prediction, grid maintenance, chatbots, safety and cybersecurity; no count published
Key findings
Exelon's 2025 Sustainability Report, released in Sep 2026, says the company built on its AI Center of Excellence in 2025 to scale AI in a structured way, ran AI literacy training for employees, and offers SPARK, an in-house AI platform on which employees develop and test ideas.
POSEIDON, an AI engine Exelon built entirely in-house to give customers specific restoration times during storms, was live at all six Exelon utilities by Dec 2025, and Exelon data scientists said in Oct 2026 that they are joining it with outage prediction and crew planning models in an Outage AI Suite and run a production-scale model that predicts underground cable failures.
Baltimore Gas and Electric has worked with Deloitte and NVIDIA since 2022 on autonomous AI drone inspections that run vision models on NVIDIA Jetson modules and train them on synthetic images from NVIDIA Omniverse, work that won an Edison Pioneers award in late 2025.
Exelon hired Tim Peterson, previously CIO and CTO of Xcel Energy, in Feb 2026 to lead a new organization joining customer experience with IT and cybersecurity, and Calvin Butler said on Oct 30, 2026 that changes made in Aug 2026 brought technology and customer operations together.
Exelon has disclosed no AI spending, AI user counts or AI-credited financial return, and Calvin Butler said on Sep 22, 2026 that he was not going to talk about how Exelon will deploy AI within its business, adding that gains in running it will come from technology partners and their use of AI.
Where Exelon sits
- Stage
- Experimenting
Exelon has reported no business return it credits to AI: its 2025 Form 10-K, 2026 proxy statement, 2025 Sustainability Report and earnings calls from Feb to Oct 2026 give no AI figure, the numbers on record are operating measures (POSEIDON restoration estimates up to three times more accurate in severe weather, drone inspection planning cut from one hour to 30 seconds), and the roughly $580 million a year in O&M savings Jeanne Jones cited in Feb 2026 is not credited to AI. Exelon has not reported a company-wide AI rollout either: SPARK, its in-house AI platform for employees, has no published user count, no executive is named as owner of AI, and CEO Calvin Butler said in Sep 2026 that he would not discuss how Exelon will deploy AI internally. Exelon is therefore at Experimenting, although POSEIDON runs in production at all six of its utilities; one return it credits to AI, or a SPARK user count with a named AI owner, would move it to Piloting.
How the stages are defined - Ownership
- No named AI owner; IT and cybersecurity under EVP Tim Peterson since Feb 2026
- Platform
- Hybrid: SPARK, POSEIDON and outage models built in-house; drone AI from Deloitte
- Models
- Not disclosed; in-house predictive models; NVIDIA-based vision models for drones
- Hosting
- Not disclosed for SPARK, POSEIDON or the outage and cable models; drone vision models run on NVIDIA Jetson modules aboard the drones; Exelon bought an NVIDIA DGX-1 system before 2023
- Reach
- POSEIDON at all six utilities; SPARK user count not disclosed
How they deploy
AI inside a reliability and affordability plan
Exelon's 2026 proxy statement says the company advanced its AI strategy in 2025, applying machine learning and generative AI to improve operational efficiency and customer interactions. Exelon's 2025 Sustainability Report (Sep 2026) lists AI in storm response, predictive maintenance from grid sensor data, analysis of utility performance data, customer service chatbots and cybersecurity, and says AI will have a profound impact on operations, customer experience and efficiency. Public Utilities Fortnightly's Aug 2026 account of Exelon's Innovation Expo names AI and automation at scale as one of four Exelon innovation priorities.
In-house models, partner-built drones
Exelon built POSEIDON, its storm restoration-time engine, entirely in-house from concept to implementation, according to the AEIC award write-up in Public Utilities Fortnightly (Dec 2025), and describes SPARK as its own AI platform. Exelon's analytics teams also built an Outage Prediction Model, a Response Planning Tool for crew deployment and underground cable failure models, which Exelon data scientists presented at industry conferences in 2026. For drone inspections, Baltimore Gas and Electric uses OptoAI, a Deloitte platform running on NVIDIA hardware, and Deloitte built the 3D models used to generate synthetic training images.
Learning from technology partners
Calvin Butler told an Atlantic Council audience on Sep 22, 2026 that he was not going to talk about how Exelon will deploy AI within its business, and said gains in running the business will come from Exelon's technology partners and their use of AI, with Exelon learning from them. Butler described the customer goal as everything on the phone with a live person when needed, and said Exelon is not there yet.
Grid orchestration
Exelon described grid orchestration in Oct 2026 as connecting drones, advanced sensors, predictive analytics and real-time grid monitoring across its utilities so that problems are found before they cause outages, an approach Laura Randle-Smith, Vice President of Exelon Infrastructure, presented in a keynote at the DistribuTECH Reliability and Resiliency conference in Aug 2026. Exelon's account of the talk does not mention AI, name models or vendors, or give figures.
Startups and research consortia as a pipeline
Exelon screens technology through a Partnership R&D Program that has funded 40 projects in nine years, research groups such as EPRI, and the Exelon Foundation's Climate Change Investment Initiative (2c2i). Exelon joined EPRI's Open Power AI Consortium at its launch with NVIDIA in Mar 2025. Through 2c2i, Salient Predictions piloted machine-learning weather forecasts with Baltimore Gas and Electric's gas supply team in 2025, and the Exelon Foundation invested in Continuum Industries, maker of the AI route-planning tool Optioneer, in Sep 2026 without saying Exelon's utilities use it.
- Deployment model
- Exelon runs AI from shared services with delivery in its utilities. Exelon Business Services Company provides IT and other support services to ComEd, PECO, BGE and Pepco Holdings at cost, and since Feb 2026 IT and cybersecurity have sat in an enterprise customer and technology organization under EVP Tim Peterson, who reports to CEO Calvin Butler. Exelon's AI Center of Excellence and Exelon Analytics data science teams, including an Infrastructure and Safety Analytics group, build models that utility teams put to work, such as Baltimore Gas and Electric's drone inspection and cable replacement planning. Exelon has not published who leads the AI Center of Excellence or where it reports.
- Model providers
- Exelon has not named a large language model provider for SPARK, its customer service chatbots or its generative AI work. Baltimore Gas and Electric's drone inspections use Deloitte's OptoAI platform, with vision models that run on NVIDIA Jetson modules aboard the drones and were trained on NVIDIA accelerated computing using synthetic images from NVIDIA Omniverse Replicator; a May 2023 NVIDIA blog by Exelon staff compared a YOLOv5 baseline with a new mask-based defect model. Exelon's outage prediction, crew planning and cable failure models are Exelon's own, and Exelon's analytics team said in Oct 2024 that it was applying multimodal vision-language models to automate engineering and design tasks.
- Data and infrastructure
- Exelon has not disclosed which clouds or data centers host SPARK, POSEIDON, its outage and cable models or its chatbots. POSEIDON combines historical and live storm restoration data to produce customer-specific restoration times, the Outage Prediction Model uses weather, grid and vegetation data, and Baltimore Gas and Electric's cable failure model draws on GIS asset attributes, environmental data and outage history and reports through a Power BI dashboard. The drone models process images at the edge on the drone instead of in the cloud and can correct GIS records, and Exelon staff wrote in May 2023 that the work built on Exelon's earlier purchase of an NVIDIA DGX-1 system. Exelon's 2025 Sustainability Report says cyber assets critical to electric and gas operations sit in restricted networks segmented from enterprise IT and the internet.
Platforms and internal tools
SPARK
SPARK is Exelon's in-house AI platform on which employees develop and test ideas, described in the 2025 Sustainability Report (Sep 2026) alongside Exelon's human-centered AI literacy training. Exelon has not said when SPARK launched, which models it uses or how many employees use it.
POSEIDON2025
POSEIDON (Personalized and Optimized Storm ETR Information Delivery and Outage Notification) is an AI analytics engine Exelon built in-house that gives customers specific estimated restoration times from the start of a storm, supporting call centers, web and mobile services and outage messaging. Exelon says POSEIDON predicts restoration times up to three times more accurately in severe weather, and it won AEIC's 2025 Achievement Award.
- Users
- All six Exelon utilities (Dec 2025)
Outage AI Suite
Outage AI Suite is the name Exelon data scientists use for joining three in-house models: the Outage Prediction Model, which forecasts outage volumes and locations days before a storm, the Response Planning Tool, which turns those forecasts into crew deployment plans, and POSEIDON. Exelon presented the suite at Utility Analytics Week in Oct 2026 and describes the work of combining the models as ongoing.
OptoAI drone inspectionWork began 2022
OptoAI is a Deloitte AI platform running on NVIDIA technology that Baltimore Gas and Electric uses for autonomous drone inspections of poles, crossarms, insulators, transformers and wires, with color-coded reports on tablets and insights on XR headsets. NVIDIA says Exelon, Deloitte and NVIDIA are working to deploy AI for field operations across Exelon's service territories.
- Models
- Vision models on NVIDIA Jetson, trained with NVIDIA Omniverse synthetic data
Vendors and partners
Deloitte2022
Supplier of the OptoAI drone analytics platform and builder of synthetic training assets · Deloitte built 3D models of grid assets and defects in Autodesk Maya and Unreal Engine for synthetic crossarm images, according to a May 2023 NVIDIA blog by Exelon staff.
NVIDIA2022
Jetson edge modules, Omniverse synthetic data and training hardware for drone inspection · Exelon staff wrote in May 2023 that the drone work built on Exelon's earlier purchase of an NVIDIA DGX-1 system.
EPRIOpen Power AI Consortium, Mar 2025
Research partner on grid-enhancing technologies, data center flexible load and power-sector AI · Exelon takes part in EPRI's GET SET program and Data Center Flexible Load Initiative (2025 Sustainability Report). Utility Dive lists Exelon among the founding members of EPRI's Open Power AI Consortium, launched with NVIDIA in Mar 2025; NVIDIA places Exelon on its executive advisory committee.
Salient Predictions2025
Machine-learning weather forecasts piloted with BGE's gas supply team · Salient Predictions is a 2c2i portfolio company; Exelon says its forecasts will inform BGE demand forecasting, procurement and storage decisions.
Leadership
AI, data and platform leaders
Executives
AI org map
AI Center of Excellence and analytics teams in shared services, with delivery in the utilities; no published AI chief
Exelon has not named an executive who owns AI. Since Feb 2026, IT and cybersecurity have sat in an enterprise customer and technology organization under EVP Tim Peterson, who reports to CEO Calvin Butler, and Exelon's 2025 Form 10-K says the Chief Information Security Officer reports to Exelon's Chief Information Officer. Calvin Butler said on Oct 30, 2026 that changes made in Aug 2026 brought technology and customer operations together, centralized engineering, design and project management, and put the utility CEOs in direct reporting lines to him.
Exelon has not published the size of its AI or analytics workforce. Public sources show an AI Center of Excellence, Exelon Analytics data science teams including an Infrastructure and Safety Analytics group, data scientists and quantitative engineers who built the outage and cable failure models, a BGE analytics and innovation team behind the drone work, and an Exelon Customer Team that described building generative AI tools in Oct 2024. A 2024 Utility Analytics Week bio of Exelon's Vice President of Innovation and Strategic Development said Exelon's Innovation Center of Excellence was then focusing on AI and generative AI. Exelon Business Services Company provides IT services to the utilities at cost, and Exelon had 20,571 employees at Dec 31, 2025.
Reporting lines
President and CEO
Customer and Technology
Tim Peterson, EVP and Chief Customer and Technology Officer
Customer strategy and experience, information technology and cybersecurity (Feb 2026); technology and customer operations combined in Aug 2026
Information Technology
Chief Information Officer, not named publicly
Line to Tim Peterson inferred from his remit, not published
Cyber Information and Security Services
Chief Information Security Officer, not named publicly
Runs enterprise-wide cybersecurity programs and gathers cyber threat intelligence (2025 Form 10-K)
Chief Operating Officer
Utility CEOs: ComEd, PECO, BGE and Pepco Holdings
Aligned directly to Calvin Butler in Aug 2026
AI Center of Excellence
Leader and reporting line not published; Exelon says it built on the center in 2025 to scale AI in a structured way
Exelon Analytics
Data science teams, including Infrastructure and Safety Analytics, behind the drone, outage and cable failure models; reporting line not published
Operation, Analytics, and Business Investments
Isaac Akridge, SVP (title as of May 2023)
Named with the drone analytics work in 2023; current structure not confirmed
Exelon's executive profiles page, checked Nov 1, 2026, lists no CIO, chief data officer or chief AI officer. Exelon's website had listed Michael Brooks, who joined in May 2024 from LKQ, as Senior Vice President and CIO; that profile page no longer exists as of Nov 1, 2026, and Exelon has not named a CIO publicly since. The line from the CIO to Tim Peterson is inferred from Peterson's IT remit, and the reporting lines of the AI Center of Excellence, Exelon Analytics and Exelon Infrastructure are not published.
Where the AI workforce sits
AI Center of Excellence and Exelon Analytics in shared services
Exelon has not published how many data scientists or AI engineers it employs.
Not disclosedUtility teams applying AI: storm response, drone inspection, cable replacement, customer operations
Exelon describes BGE's analytics and drone teams, cable planners and utility storm teams as users of AI tools; no AI headcount is given.
Not disclosed
Named people
Use cases
Exelon names AI in storm restoration estimates (POSEIDON), outage prediction and crew planning, underground cable failure prediction, drone inspections at Baltimore Gas and Electric, predictive maintenance from grid sensor data, analysis of utility performance data, customer service chatbots, safety investigations and cybersecurity, and has not published a count of its AI use cases. Exelon has reported no business-level return credited to AI for any of them and no firm-level AI return; the $580 million in annual O&M savings against inflation-level growth that Jeanne Jones cited in Feb 2026 is not attributed to AI.
| Use case | Area | Status | ROI |
|---|---|---|---|
| POSEIDON storm restoration estimates POSEIDON gives customers specific estimated restoration times from the start of a storm by combining historical and live restoration data, and supports call centers, web and mobile services and outage messaging at all six Exelon utilities. | Storm response and customer communication | In production | |
| Outage prediction and crew planning Exelon's Outage Prediction Model forecasts outage volumes and locations days before a storm from weather, grid and vegetation data, and its Response Planning Tool turns those forecasts into crew deployment plans; Exelon data scientists describe joining both with POSEIDON in an Outage AI Suite, and Exelon has not said whether the two models run in production. | Storm response | Pilot | |
| Underground cable failure prediction Exelon models built from outage history, cable attributes and environmental conditions estimate the likelihood of underground residential distribution cable faults within the next year, feeding planner dashboards for replacement and capital decisions; the work began at PECO and Pepco Holdings, reached BGE by May 2026, and Exelon described it in Oct 2026 as a production-scale implementation across several operating companies. | Grid operations and asset management | In production | |
| Autonomous AI drone inspections Baltimore Gas and Electric's drones fly autonomously, detect defects in poles, crossarms, insulators and transformers with vision models running on board, and correct GIS asset records, using Deloitte's OptoAI platform on NVIDIA hardware. | Grid operations (BGE) | In production | |
| Predictive maintenance from grid sensors Exelon applies AI and machine learning to data from sensors on the grid to detect likely equipment failures before they occur. | Grid operations | In production | |
| AI chatbots for customers Exelon says it provides personalized assistance to customers aided by AI chatbots; Exelon has not named the chatbots, the channels or the models behind them. | Customer service | In production | |
| Generative AI for representatives Exelon's Customer Team described developing generative AI tools to train customer service representatives, help draft replies to customer emails and answer internal company questions in Oct 2024; Exelon has not said whether they went live. | Customer service | Pilot | |
| ACE Navigator and public safety analytics An Exelon data scientist described in Sep 2026 ACE Navigator, an AI-assisted tool that supports investigations of safety events and corrective actions, and a generative AI process that enriches historical public safety data for analytics and reporting, both with a human in the loop; Exelon has not said whether they run in production. | Safety | Pilot | |
| Multimodal models for engineering and design tasks Exelon's analytics team presented case studies in Oct 2024 on using multimodal vision-language models to automate time-intensive engineering and design tasks. | Engineering | Pilot | |
| AI-assisted performance analysis Exelon's data analysts use AI to interpret large volumes of utility performance data. | Analytics | In production | |
| AI in cyber defense Exelon uses AI systems in cybersecurity to protect utility networks and customer data. | Cybersecurity | In production | |
| Machine-learning weather forecasts Salient Predictions, a 2c2i portfolio company, piloted machine-learning sub-seasonal and seasonal weather forecasts with BGE's gas supply team in 2025 to inform demand forecasting, procurement and storage. | Gas supply (BGE) | Pilot |
Adoption
Exelon's earliest public AI work sits in analytics: Exelon bought an NVIDIA DGX-1 system, Baltimore Gas and Electric began AI drone inspections with Deloitte and NVIDIA in 2022, and Exelon's analytics and customer teams were testing generative AI and multimodal models by Oct 2024. Exelon launched POSEIDON in 2025 and had it live at all six utilities by Dec 2025, its 2025 Sustainability Report (Sep 2026) describes an AI Center of Excellence, AI literacy training and SPARK, an in-house platform for employees, and its data scientists described an underground cable failure model running at production scale across several utilities in Oct 2026. Exelon has not published how many employees use SPARK or any other AI tool.
By the numbers
Rollout, first pilot to today
2022
Baltimore Gas and Electric began working with Deloitte and NVIDIA on autonomous AI drone inspection of grid assets.
May 2023
Exelon analytics staff described training drone defect-detection models on synthetic crossarm images from NVIDIA Omniverse Replicator, with validation by business stakeholders under way.
Oct 2024
Exelon presented generative AI work at Utility Analytics Week: multimodal models for engineering automation, customer tools for representative training, email replies and internal questions, and a generative AI governance and implementation plan.
2025
Exelon launched POSEIDON, its AI engine for customer storm restoration estimates, and built on its AI Center of Excellence to scale AI in a structured way.
Mar 2025
Exelon joined EPRI's Open Power AI Consortium, launched with NVIDIA to build open AI models for the power sector.
Nov 2025
Exelon and BGE's AI drone inspection work with Deloitte and NVIDIA won Public Utilities Fortnightly's Edison Pioneers award for collaboration in innovation.
Dec 2025
Public Utilities Fortnightly reported that POSEIDON, winner of AEIC's 2025 Achievement Award, was live at all six Exelon utilities.
Feb 2026
Tim Peterson joined Exelon to lead a new organization combining customer strategy and experience with IT and cybersecurity.
May 2026
Exelon and BGE staff presented a machine learning model that predicts underground cable faults to prioritize BGE cable replacements, building on earlier work at PECO and Pepco Holdings.
Sep 2026
Exelon's 2025 Sustainability Report described SPARK, its in-house AI platform for employees, and human-centered AI literacy training.
Oct 2026
Exelon data scientists described the underground cable failure model as a production-scale implementation across several operating companies and presented the Outage AI Suite at Utility Analytics Week.
Economics & controls
Spend and ROI
Technology budget
Exelon does not disclose an IT or AI budget. Exelon's capital plan stood at $41.7 billion through 2029 in Oct 2026, and Exelon expected to deploy about $10 billion of capital in 2026; Exelon Business Services Company charges IT and other support services to the utilities at cost. Exelon's 2025 Form 10-K says software development costs, including those for cloud-based solutions treated as service arrangements, are capitalized and amortized once the projects are in use.
What AI itself costs
Exelon has not disclosed what it spends on AI models, SPARK, POSEIDON, the outage and cable models, the drone program or its partners Deloitte and NVIDIA.
Reported ROI
Exelon has reported no financial return credited to AI. Exelon says POSEIDON predicts restoration times up to three times more accurately in severe weather, and Exelon said in Dec 2025 that drone inspection planning time fell from one hour to 30 seconds and that weeks of manual photo sorting were eliminated; NVIDIA's case study describes drone operation time falling from up to an hour to as little as 30 seconds. None of these figures is tied to cost, revenue or headcount. NVIDIA says the drone work is set to decrease operational costs, a projection with no figure, and Exelon's Oct 2026 Utility Analytics Week session on the cable failure model listed measuring its operational and financial benefits as a topic without publishing a figure.
Cost programs
Jeanne Jones said on Feb 12, 2026 that Exelon's cost focus saves about $580 million in O&M a year compared with inflation-level growth over the past decade, and that Exelon expected nearly flat expense growth from 2024 to 2026. Calvin Butler said in Sep 2026 that Exelon would have taken $1 billion of O&M costs out of its system by the end of 2026. Exelon's Q3 2026 Form 10-Q books third-party consulting and severance costs for a cost management program, $57 million after tax in the first nine months of 2026. Exelon attributes none of these savings to AI.
Headcount
Exelon had 20,571 employees at Dec 31, 2025, including 6,688 at ComEd, 3,383 at BGE, 3,169 at PECO and 4,422 at Pepco Holdings, and 8,656 were covered by collective bargaining agreements. Exelon has made no public statement linking AI to headcount.
Governance and risk
Board oversight
Exelon's 2026 proxy statement assigns oversight of operational technology, cybersecurity and physical security risk to the board's Operations, Safety and Customer Experience Committee, with the Chief Operating Officer and the Chief Technology and Customer Officer as executive leads. Exelon's proxy says the board recently added a director with expertise in cybersecurity, technology and privacy, and lists technology and innovation among the skills it seeks.
AI Center of Excellence and literacy training
Exelon's 2025 Sustainability Report says the AI Center of Excellence gives Exelon a structured approach to scale AI responsibly, and that human-centered AI literacy training teaches employees to use AI safely. Exelon's materiality assessment in the same report says AI needs careful management because the technology is at a relatively nascent stage and carries potential for unintended impacts or misuse. Exelon's Vice President of Innovation and Strategic Development presented a generative AI governance and implementation plan at Utility Analytics Week in Oct 2024, and Exelon has not published the plan.
Generative AI risks
Exelon's Customer Team said in Oct 2024 that minimizing risk matters for its generative AI tools because they can carry bias and invent information, and that many analytics models never reach production. An Exelon data scientist described a human-in-the-loop approach to generative AI in safety operations in Sep 2026. Exelon's filings describe cybersecurity controls in detail but do not set out a model risk process for AI.
Cybersecurity and grid isolation
Exelon's 2025 Form 10-K warns that the rapid evolution and adoption of AI may intensify its cybersecurity risks, and says the Chief Information Security Officer, who reports to the Chief Information Officer, manages enterprise-wide cybersecurity programs through Cyber Information and Security Services. Exelon's 2025 Sustainability Report says cyber assets critical to electric and gas operations are isolated from the enterprise IT network and the internet and monitored continuously, and that Exelon's security controls align with the NIST Cybersecurity Framework.
Recent signals
Oct 30, 2026
Calvin Butler said on Exelon's Q3 2026 call that changes made in Aug 2026 brought technology and customer operations together, centralized engineering, design and project management, and put the utility CEOs in direct reporting lines to him.
Oct 22, 2026
Exelon data scientists told Utility Analytics Week about a production-scale AI model that predicts underground residential cable failures across several Exelon operating companies, a day after presenting Exelon's Outage AI Suite for storm prediction, staffing and restoration.
Sep 22, 2026
Calvin Butler told an Atlantic Council event that he was not going to talk about how Exelon will deploy AI within its business and that gains in managing the business will come from technology partners and their use of AI.
Sep 22, 2026
An Exelon data scientist presented ACE Navigator, an AI-assisted tool for investigating safety events and corrective actions, and a generative AI process for enriching historical public safety data, both with a human in the loop.
Sep 16, 2026
The Exelon Foundation added Continuum Industries, whose Optioneer platform uses AI to generate and compare infrastructure routes, to its 2c2i startup portfolio; Exelon did not say its utilities use the tool.
Sep 2026
Exelon's 2025 Sustainability Report said Exelon built on its AI Center of Excellence in 2025, runs AI literacy training, offers SPARK, an in-house AI platform for employees, and uses AI chatbots in customer service.
Aug 25, 2026
Exelon announced that COO Mike Innocenzo, one of the two executive leads for the board committee that oversees operational technology and cybersecurity risk, will leave in 2027, that CFO Jeanne Jones will become Executive Vice President of Finance and Strategy, and that Robert Kleczynski will become CFO, both on Oct 5, 2026.
Aug 2026
Public Utilities Fortnightly's account of Exelon's Innovation Expo listed AI and automation at scale among Exelon's four innovation priorities, alongside affordability, reliability and the future workforce.
Jul 30, 2026
Exelon cut its high-probability data center pipeline to 11 gigawatts from 18 gigawatts, with Jeanne Jones saying speculative projects had been weeded out, as AI data center plans faced pushback, Bloomberg reported.
May 20, 2026
Exelon and BGE staff presented a machine learning model that estimates the likelihood of underground cable faults within a year to prioritize BGE cable replacements, reported through a Power BI dashboard.
Mar 18, 2026
Exelon's 2026 proxy statement said the company advanced its AI strategy, applying machine learning and generative AI to operational efficiency and customer interactions, and launched POSEIDON.
Feb 12, 2026
Exelon's 2025 Form 10-K warned that the rapid adoption of AI may intensify its cybersecurity risks, and Jeanne Jones cited about $580 million in annual O&M savings against inflation-level growth without attributing them to AI.
Jan 8, 2026
Exelon named Tim Peterson, CIO and CTO at Xcel Energy, Executive Vice President and Chief Customer and Technology Officer, effective Feb 2026, leading a new organization that combines customer experience with IT.
Dec 3, 2025
Exelon reported that its and BGE's AI drone inspection work with Deloitte and NVIDIA won the Edison Pioneers award for collaboration in innovation.
Dec 2025
Public Utilities Fortnightly's write-up of AEIC's 2025 Achievement Awards said POSEIDON was developed entirely in-house and is live at all six Exelon utilities.
Quick answers
- How does Exelon deploy AI?
- Exelon deploys AI through an AI Center of Excellence, SPARK, an in-house platform on which employees develop and test AI ideas, and models its analytics teams build for its utilities, led by POSEIDON storm restoration estimates and an underground cable failure model. Baltimore Gas and Electric runs AI drone inspections built with Deloitte and NVIDIA. Exelon has published no AI spending, usage counts or AI-credited returns, and CEO Calvin Butler said in Sep 2026 he would not discuss internal AI plans.
- What does Exelon use AI for?
- Exelon names AI in storm restoration estimates (POSEIDON), outage prediction and crew planning, underground cable failure prediction, drone inspections at Baltimore Gas and Electric, predictive maintenance from grid sensor data, analysis of utility performance data, customer service chatbots, safety investigations and cybersecurity, and has not published a count of its AI use cases. Exelon has reported no business-level return credited to AI for any of them and no firm-level AI return; the $580 million in annual O&M savings against inflation-level growth that Jeanne Jones cited in Feb 2026 is not attributed to AI.
- Who leads AI at Exelon?
- AI at Exelon is led by Tim Peterson (Executive Vice President and Chief Customer and Technology Officer). Ownership of AI on the org chart: No named AI owner; IT and cybersecurity under EVP Tim Peterson since Feb 2026.
- What AI platform does Exelon use?
- SPARK, an in-house AI platform, lets Exelon employees develop and test AI ideas, and Exelon says it built on its AI Center of Excellence in 2025 to scale AI in a more structured way (2025 Sustainability Report, Sep 2026).
- Which AI models does Exelon use?
- Exelon has not named the models behind SPARK or POSEIDON, and its analytics teams build their own outage and cable failure models; Baltimore Gas and Electric's drone inspections run vision models on NVIDIA Jetson modules through Deloitte's OptoAI platform, trained with NVIDIA Omniverse synthetic data.
- How many Exelon employees use AI?
- POSEIDON runs at all six Exelon utilities (Dec 2025), but Exelon has not said how many of its 20,571 employees use SPARK or any other AI tool.
- How far has AI spread at Exelon?
- Exelon has reported no business return it credits to AI: its 2025 Form 10-K, 2026 proxy statement, 2025 Sustainability Report and earnings calls from Feb to Oct 2026 give no AI figure, the numbers on record are operating measures (POSEIDON restoration estimates up to three times more accurate in severe weather, drone inspection planning cut from one hour to 30 seconds), and the roughly $580 million a year in O&M savings Jeanne Jones cited in Feb 2026 is not credited to AI. Exelon has not reported a company-wide AI rollout either: SPARK, its in-house AI platform for employees, has no published user count, no executive is named as owner of AI, and CEO Calvin Butler said in Sep 2026 that he would not discuss how Exelon will deploy AI internally. Exelon is therefore at Experimenting, although POSEIDON runs in production at all six of its utilities; one return it credits to AI, or a SPARK user count with a named AI owner, would move it to Piloting.
Open questions
- Who leads Exelon's AI Center of Excellence, is it the Innovation Center of Excellence that Exelon described in 2024 as focused on AI and generative AI, and does it report to Tim Peterson's customer and technology organization?
- Which models does SPARK use, where is it hosted, and how many of Exelon's 20,571 employees use it?
- Which clouds or data centers run POSEIDON, SPARK, the outage and cable models and Exelon's customer chatbots?
- What business return do POSEIDON, the BGE drone inspections, the cable failure model, the chatbots and predictive maintenance deliver, in call volumes, O&M cost or outage minutes?
- How much does Exelon spend on AI, and what share of its IT spending does AI take?
- How many AI use cases does Exelon run in production, and are the Outage Prediction Model, the Response Planning Tool and ACE Navigator among them?
- Did the generative AI tools Exelon's Customer Team described in Oct 2024, for representative training, email replies and internal questions, reach production?
- How far have AI drone inspections spread beyond BGE to ComEd, PECO and Pepco Holdings?
- Does any of the $580 million in annual O&M savings or the 2026 cost management program come from AI or automation?
- Who is Exelon's Chief Information Officer now that the profile of Michael Brooks, the CIO hired in 2024, is gone from Exelon's site?
- Does Exelon have a written AI policy or model risk process covering SPARK, POSEIDON and its customer chatbots, and what does its 2024 generative AI governance plan contain?
Sources
Exelon
- Executive profiles
- Calvin Butler, President and CEO, profile
- Tim Peterson, EVP and Chief Customer and Technology Officer, profile
- Colette Honorable, EVP and Chief Legal Officer, profile
- Exelon names Tim Peterson Chief Customer and Technology Officer (Jan 8, 2026)
- Exelon Sustainability Report 2025 with appendix (Sep 2026)
- Exelon leads the industry with three top innovator awards (Dec 3, 2025)
- How Exelon is preventing outages before they happen (Oct 1, 2026)
- How Exelon is preparing the grid for AI and data center growth (Oct 9, 2026)
- Exelon adds Continuum Industries to 2c2i portfolio (Sep 16, 2026)
- Exelon reports third quarter 2026 results (Oct 30, 2026)
Filings and investor materials
- Exelon 2025 Form 10-K (filed Feb 12, 2026)
- Exelon 2026 proxy statement (Mar 18, 2026)
- Exelon Q3 2026 Form 10-Q (Oct 30, 2026)
- Exelon announces key leadership position changes, 8-K exhibit (Aug 25, 2026)
- Q3 2026 earnings call transcript, The Motley Fool (Oct 30, 2026)
- Q2 2026 earnings call transcript, The Motley Fool (Jul 30, 2026)
- Q4 2025 earnings call transcript, The Motley Fool (Feb 12, 2026)
Partners and industry bodies
- NVIDIA case study: Exelon autonomous drone grid inspection
- NVIDIA Technical Blog: Exelon uses synthetic data generation to automate drone inspection (May 1, 2023)
- Deloitte: AI field solutions for drone power line inspection
- AEIC 2025 Achievement Awards, Public Utilities Fortnightly reprint (Dec 2025)
- AEIC Achievement Awards winners
- NVIDIA: EPRI, NVIDIA and collaborators launch Open Power AI Consortium (Mar 20, 2025)
Conference sessions
- Utility Analytics Week 2026 agenda: Exelon's Outage AI Suite (Oct 21, 2026)
- Utility Analytics Week 2026: Predicting URD cable failures with AI across Exelon operating companies (Oct 22, 2026)
- Utility Analytics Institute: Exelon's human-in-the-loop approach to generative AI (Sep 22, 2026)
- Utility Analytics Institute: URD failure prediction at BGE (May 20, 2026)
- DTECH 2027 session listing: Inside Exelon's AI-driven storm outage suite
- Utility Analytics Week 2024: Exelon overcomes generative AI obstacles in customer data science (Oct 16, 2024)
- Utility Analytics Week 2024: Exelon advances engineering automation with generative AI (Oct 16, 2024)
- Utility Analytics Week 2024: speaker bio, Exelon Vice President of Innovation and Strategic Development
News and events
- Public Utilities Fortnightly: Exelon, AI at the edge for drone autonomous inspections (Nov 2025)
- Public Utilities Fortnightly: Innovation in Action, Exelon's Innovation Expo (Aug 2026)
- Utility Dive: EPRI launches consortium to develop power sector AI applications (Mar 21, 2025)
- Atlantic Council: Exelon CEO Calvin Butler on powering the AI future (Sep 22, 2026)
- Bloomberg via Energy Connects: Exelon's data center pipeline cut as AI plans face pushback (Jul 30, 2026)
Data confidence
- Confirmed
- Stated in a primary or reputable public source.
- Estimated
- Derived or approximated from public reporting.
- Unconfirmed
- Reported, but not yet corroborated.
Exelon profile · Enterprise AI diffusion report by Imagine · facts checked against their sources on Nov 1, 2026
