Type to search every page. Use the arrow keys to move through results.
Industries
Leaders in oil and gas, refining, metals, power, and renewable energy answer for assets that must run safely, stay protected and earn their cost of capital. Published research points to data, skills and the protection of control systems as what decides whether AI can be used on them.
In Canada, manufacturing sits just below the all-industry figure for planned use of AI, and mining and oil and gas well below it. In Statistics Canada’s survey for the third quarter of 2026, 14.8% of businesses in mining, quarrying, and oil and gas extraction expected to use AI over the next 12 months, as did 24.4% of manufacturers. The figure for all businesses was 25.2%. A year earlier the manufacturing figure was 7.2%.
What leaders in these industries answer for is output that stays safe, protected and running, so much of the published evidence concerns the conditions around AI. The International Energy Agency (IEA) lists lack of access to data, concerns about systems that do not work together and critical gaps in skills among the limits on using AI in energy. In May 2025 US government agencies warned that control systems in oil and natural gas were being targeted.
The five sub-industries run different assets, yet the research points to the same questions in each. Can the data in operating systems be used? Are there people to run a tool and question it? Is what the tool connects to protected? We start with those questions. We help you decide which uses are worth the capital, train your people, help you hire the skills you lack, and set the governance and security around each tool.
24
supercomputers run by oil and gas companies were among the world’s 500 fastest in 2024, up from 11 in 2000, according to the International Energy Agency
0.5%
is the American Chemistry Council’s forecast rise in US chemical output volumes in 2026, with only 12 of 20 key chemistry-consuming industries expected to expand
65,712
is the gap between the hires that Canadian mining as a whole, including oil sands mining and mining support services, needs from 2026 to 2035 and the entrants expected, in the baseline forecast of the Mining Industry Human Resources Council
13 of 23
assessment areas of the North American power system face resource adequacy challenges over the next 10 years, according to NERC’s 2025 long-term assessment
North American Electric Reliability Corporation (NERC), 2025
2,500 GW
and more of renewable, large-load and storage projects are stalled in grid connection queues worldwide, according to the International Energy Agency
The figures on this page are third-party research. Each links to the organization that published it, and none describes our own work.
Statistics Canada asked businesses across industries about their plans for AI in the third quarters of 2025 and 2026. Planned use in manufacturing rose more than threefold. In mining and oil and gas it rose from 3.4% to 14.8%. Both remain below the all-industry figure.
These figures are plans, and the survey table does not measure use. The table covers 16 industries, all shown here.
Statistics Canada, 2026.
Leaders in oil and gas answer for how capital is spent, how much methane their operations release and whether their control systems stay protected.
The industry has long worked with large volumes of data. The International Energy Agency counts 24 supercomputers run by oil and gas companies among the world’s 500 fastest in 2024, up from 11 in 2000, and calls oil and gas an early adopter of AI.
In each of the Canada Energy Regulator’s four scenarios, natural gas production in 2050 is above the record 18.3 billion cubic feet a day reached in 2024.
The 2050 values are projections from the regulator’s four scenarios. The 2024 value is the level reported for that year.
Canada Energy Regulator, 2026.
What leaders answer for, and our part
Capital is limited, and leaders are asked for more from the assets they already run. A leader answers for which technology spending earns its place in that budget.
Our part
We work out what each technology investment would return against the others competing for the same capital, and which are worth making.
Technology strategyLeaders answer for the methane their operations release and for showing that leaks are found and repaired. Leak detection is among the uses of AI the IEA lists for oil and gas.
The IEA estimates that oil operations released 45 million tonnes of methane in 2025 and natural gas operations 36 million tonnes, with upstream activities causing 80% of oil and gas methane. It says three-quarters of oil and gas methane emissions can be abated with existing technology. Source: International Energy Agency
Our part
We build automations and AI agents that bring leak readings together for your operations teams and raise what needs a response, with your people deciding what to do.
Advanced automations and AI agentsThe systems that run wells and plants have become targets, and each AI tool connected to them adds another point that must be protected. That second point is our own reading, since the agencies’ guidance concerns control systems in general.
In May 2025 the US Cybersecurity and Infrastructure Security Agency (CISA), the Federal Bureau of Investigation (FBI), the Environmental Protection Agency and the Department of Energy set out five measures for organizations that run control systems. They include removing connections to the public internet, changing default passwords, securing remote access, separating business and operating networks, and keeping the ability to run manually through tested continuity plans. Source: CISA, FBI, EPA and US Department of Energy
Our part
We set what an AI tool may connect to and what it may do there, with security controls in place before the connection is made.
Governance, compliance and securityWith capital held tight, leaders are asked for more output and uptime from the assets already running. The question is which uses of AI would repay the effort on their own assets.
The IEA lists subsurface data processing, reservoir simulation, remote operations, predictive maintenance, regulatory compliance, leak detection and rapid automation as uses of AI in oil and gas. Its list is unranked and carries no figures. Source: International Energy Agency
Our part
We study how your operations work, in person, at your sites and offices, and develop a Consier AI solution from what we find.
Consier AIEngineers, geoscientists and operations leads are asked to stand behind what a tool tells them about reservoirs, equipment and emissions. They can do that only if they understand a tool well enough to question it.
The IEA says AI-related skills are much less prevalent in the energy sector than in other sectors. Source: International Energy Agency
Our part
We train your engineers and operations teams to use and question AI tools, and help you hire the skills you lack.
Workforce trainingWhat holds organizations back
The questions we start with
Leaders in refining and petrochemicals answer for the safe running of plants that operate around the clock and for what those plants earn. The US Energy Information Administration reports that US operable refining capacity fell by about 1% during 2025, to 18.2 million barrels a day on 1 January 2026.
The American Chemistry Council forecasts that US chemical output volumes will rise only 0.5% in 2026 and 1.5% in 2027, and that only 12 of 20 key chemistry-consuming industries will expand in 2026.
In a survey of oil and gas executives, 41% of their organizations apply AI in refining and a further 52% expect to do so within three years.
41% apply AI in refining
52% expect to apply AI in refining within three years
The survey covers oil and gas organizations, so it does not speak for refiners alone. It reports use and expectation. It does not measure results. The figures are those of a technology company’s research institute.
What leaders answer for, and our part
A leader has to show that a tool fits the plant’s safety management before it reaches a running unit.
The US Chemical Safety and Hazard Investigation Board’s February 2026 volume of incident reports describes serious chemical industry accidents linked to ineffective safety management when equipment is opened. They include a refinery fire in California on 1 February 2025 with about US$924 million in property damage. Source: US Chemical Safety and Hazard Investigation Board
Our part
We help you test each proposed AI use against your safety management before it reaches a running unit, and settle who may approve it.
Governance, compliance and securityWith capacity closing in some regions and little growth in chemical output, leaders are asked to make the plants they have earn more. Which uses of AI return enough on a given plant is a question to settle plant by plant.
The US Energy Information Administration reports that two US refineries closed in 2025, together about 400,000 barrels a day, and that the West Coast lost about 5% of its refining capacity. Source: US Energy Information Administration
Our part
We estimate what each AI use would return on a given plant, so that the uses you pursue are those that repay the effort.
Technology strategyUtilization compares what a plant produces with the most it could. For plants that run around the clock it decides what each unit earns, so leaders answer for how much capacity is used and for what keeps it from being used.
The Chemistry Industry Association of Canada reports that capacity utilization in Canada’s chemical and plastic product sectors has not recovered since a sharp decline in 2018 and 2019. Source: Chemistry Industry Association of Canada
Our part
We analyze your plant and maintenance data to show where capacity is lost, and build the regular reports that keep those losses in view.
Data analytics, reporting and dashboardsPlant systems link controls, maintenance and production data. Connecting that data to AI tools widens what has to be protected. That is our own reading, since the assessment below concerns critical infrastructure in general.
The Canadian Centre for Cyber Security calls ransomware the top cybercrime threat to Canada’s critical infrastructure and reports that ransomware incidents have grown on average 26% year over year since 2021. It says some pro-Russia non-state actors have tried to compromise operational technology in critical infrastructure. Source: Canadian Centre for Cyber Security
Our part
We set what plant data an AI tool may reach, and prepare your systems to be recovered if they are disrupted.
Local hosting and emergency readinessWhat holds organizations back
The questions we start with
Leaders in metals and heavy manufacturing answer for output and cost, for safety on site and for a skilled workforce that is hard to replace. In its baseline forecast, Canada’s Mining Industry Human Resources Council expects Canadian mining to need 246,422 hires from 2026 to 2035, against expected entries of 180,710.
The World Steel Association reports that world crude steel production in 2025 was 1,849.4 million tonnes, and that the 70 countries reporting to it produced 2.0% less than in 2024.
Of the 246,422 hires that Canadian mining needs from 2026 to 2035 in its baseline forecast, 212,556 replace people leaving the workforce or the industry.
The width of each part shows its share of the 246,422 total. The council’s definition of mining includes extraction and milling, oil sands mining, mining support services and primary metal manufacturing.
What leaders answer for, and our part
Most of the hiring that Canadian mining expects to need replaces people who are leaving, so leaders answer for how skills are passed on and where new ones come from.
In Statistics Canada’s spring 2026 survey, 12.0% of manufacturing businesses named a lack of skilled workers as a barrier that limits their use of AI. Source: Statistics Canada
Our part
We train your people to work with analytics and automation, and help you hire the skills you lack.
Workforce trainingPreventing deaths on site is the first test of a mining leader’s record. Any tool proposed for safety has to be judged against the hazards the industry reports, such as fall of ground and mobile equipment, before it is relied on.
ICMM reports that 39 people from its 26 member companies lost their lives at work in 2025, against 43 in 2024 and 35 in 2023. Fall of ground and mobile equipment each caused 10 of the deaths, and ICMM describes the trend since 2021 as deeply concerning. Source: International Council on Mining and Metals (ICMM)
Our part
We help you set how a safety tool will be judged, by which measures and against which hazards, before you buy one.
Technology strategyWith reported steel output lower in 2025, leaders answer for cost per tonne and for equipment that does not stop the line. Manufacturers report their use of AI in those same places.
The US National Institute of Standards and Technology reports figures it attributes to the Manufacturing Leadership Council on where manufacturers use AI today. Preventative or predictive maintenance and process improvement are each at 54%, productivity and cost reduction at 50% and quality improvement at 49%. Source: US National Institute of Standards and Technology, Manufacturing Extension Partnership
Our part
We build automations and AI agents for the maintenance and process work you choose, and agree beforehand how the return will be measured.
Advanced automations and AI agentsEnergy use and emissions are board matters for producers, and a measured baseline comes before any claim about what a tool could change.
The World Steel Association says that in 2024 each tonne of steel produced led, on average, to 2.18 tonnes of greenhouse gas emissions, counted as carbon dioxide equivalent, direct and indirect. It says steel emissions represent 7% to 8% of global greenhouse gas emissions from human activity. Source: World Steel Association (worldsteel)
Our part
We analyze your production and energy data and build the regular reports that give your board a measured baseline.
Data analytics, reporting and dashboardsMoving a pilot to use across a whole facility needs an owner, a budget and a measure of what it returns.
Deloitte surveyed 600 executives at large US manufacturers in August and September 2024 for its 2025 report. Of them, 29% said they use AI at the facility or network level and 23% were piloting it. Source: Deloitte
Our part
We help you choose which pilots to carry across a facility, name an owner for each and settle how its return will be measured.
Technology strategyWhat holds organizations back
The questions we start with
Leaders of electric and water utilities answer to regulators and the public for supply that is adequate, networks that stay reliable and capital spent where it is needed. Demand growth is now testing all three.
In Canada, the Canada Energy Regulator projects electricity demand rising by 26% to 85% from 2023 to 2050 across its scenarios. Its Higher scenario reflects greater data centre load growth and higher assumed economic growth.
Of the three uses charted, the grid operators surveyed most often use AI for maintenance and operation planning, and least often in real-time operations.
The IEA words the first two as “implemented AI for” and the last as “use AI for”. The first is printed as “nearly 70%”. The categories are the IEA’s and overlap. Where it reports these figures, the IEA does not say how many operators were surveyed.
International Energy Agency, 2025.
What leaders answer for, and our part
Regulators hold utilities to supply that is adequate, and the demand they plan for keeps rising. AI is already used in the planning that comes first: in the International Energy Agency’s survey of grid operators, 54% had implemented it for grid development planning.
The International Energy Agency reports that over 2,500 gigawatts of renewable, large-load and storage projects are stalled in grid connection queues worldwide. Source: International Energy Agency
Our part
We sit with your planning teams to see which forecasting and scheduling tools they rely on, and help you decide which are worth extending.
Technology strategyA utility answers for what it commits to build for a new data centre. NERC says that projections for new resources and transmission lag what new data centres and other large loads require.
A 2024 report from the Lawrence Berkeley National Laboratory, published by the US Department of Energy, estimates that US data centres used 176 terawatt hours of electricity in 2023, 4.4% of US electricity. It projects 325 to 580 terawatt hours in 2028, 6.7% to 12%. Source: US Department of Energy (reporting the Lawrence Berkeley National Laboratory’s 2024 report)
Our part
We build the reports and dashboards that bring demand forecasts, connection requests and capacity together for your executives and your regulator.
Data analytics, reporting and dashboardsReliability is the measure regulators apply to a utility, so leaders are most careful where a tool would act on the live network. Fewer than a quarter of the grid operators the International Energy Agency surveyed use AI there.
The US Department of Energy’s 2024 report says grid operators are likely to be very reluctant to adopt new technologies without robust assurance that they will not compromise reliability. Source: US Department of Energy
Our part
We help you set which decisions a tool may inform and which it may make, who approves each, and how that is recorded in your governance.
Governance, compliance and securityUtilities answer for networks that need renewing and for how the work is paid for. In the American Water Works Association’s 2026 survey of 2,171 water professionals, infrastructure renewal and replacement was the top challenge for water utilities, followed by financing infrastructure upgrades.
The US Environmental Protection Agency says national studies find that on average 14% of the water treated by water systems is lost to leaks, and some systems have reported losses above 60%. Source: US Environmental Protection Agency
Our part
We analyze in person how your engineering, maintenance and finance teams work today, and build a Consier AI solution around that.
Consier AIA board has to show that AI in systems the public depends on is overseen and secure, and the people who run those systems must be able to question it.
The American Water Works Association’s 2026 summary says water utilities are cautiously exploring new technologies like AI, recognizing both their potential benefits and the associated risks, especially in cybersecurity. Source: American Water Works Association (AWWA)
Our part
We train your engineers and operators to use and question AI tools, and help you hire the skills that bridge power systems and AI.
Workforce trainingWhat holds organizations back
The questions we start with
Leaders in renewable energy answer for output that can be connected and sold as much as for output that can be built. The IEA reports that grid infrastructure takes 5 to 15 years to plan and complete, against 1 to 5 years for renewable projects.
The US Energy Information Administration reports that California’s grid operator curtailed, or held back, 3.4 million megawatt-hours of utility-scale wind and solar output in 2024, 29% more than in 2023. Solar was 93% of the energy curtailed.
At the end of 2024, 956 gigawatts of solar and 890 gigawatts of storage were waiting in US grid connection queues, against 271 of wind and 136 of natural gas.
The queues are those of over 50 transmission operators, which together cover about 97% of US generating capacity. Not every place in a queue becomes a project that is built.
Lawrence Berkeley National Laboratory (US Department of Energy), via OSTI, 2024.
What leaders answer for, and our part
A project that cannot connect earns nothing, and the grid operator sets much of the wait. Leaders answer for which projects are worth carrying through that wait. The IEA says lower-risk tools for forecasting, maintenance, inspection and planning are scaling first in the grid.
Lawrence Berkeley National Laboratory reports that in the US the median time from grid connection request to commercial operation was under two years for projects built from 2000 to 2007. For those built from 2018 to 2024 it was over four years. Of the capacity requested from 2000 to 2019, 13% reached commercial operation and 77% was withdrawn by the end of 2024. Source: Lawrence Berkeley National Laboratory (US Department of Energy), via OSTI
Our part
We bring the queue position, studies and cost of each project into one view, so that your leaders can see which are worth pursuing.
Data analytics, reporting and dashboardsVariable output has to be sold ahead of time, and a poor forecast means more reserves held and more output wasted. The IEA lists an example of AI-assisted wind forecasting in the US with forecast error cut from 16.8% to 10.1% and US$49 million saved over five years.
IRENA cites examples from Australia, India and the United Kingdom in which AI-enhanced forecasting gave forecasts up to 45% more accurate than traditional methods and reduced curtailment. IRENA takes these figures from vendor and development-agency sources. Source: International Renewable Energy Agency
Our part
We build automations and AI agents that bring weather, output and market data together for your operators, with your people making each call.
Advanced automations and AI agentsInvestors and lenders examine the climate figures a company reports, and the rules are still moving. The IFRS Foundation says more than 40 jurisdictions, Canada among them, have decided to use or are taking steps to introduce the standards of the International Sustainability Standards Board.
On 23 April 2025 the Canadian Securities Administrators said they were pausing work on a new mandatory climate-related disclosure rule, and that securities legislation already requires issuers to disclose material climate-related risks. Source: Canadian Securities Administrators (release posted by the Alberta Securities Commission)
Our part
We set up governance and controls over the data behind each reported figure, so that you can trace it to its source.
Governance, compliance and securityOperators and developers are asked to rely on forecasts and alerts from tools they did not build, which requires enough people who can judge the output.
In IRENA’s survey, with 61 responses mostly from government agencies and utilities, 60% of respondents called a shortage of skilled professionals a critical or high barrier to digitalizing power systems. Source: International Renewable Energy Agency
Our part
We train your operators and analysts to use and question digital tools, and help you hire the skills you lack.
Workforce trainingWhat holds organizations back
The questions we start with
These are problems in the sectors above that AI is being used on today. For each, we set out what published evidence shows and what we would do.
The problem
The North Sea Transition Authority reports that production efficiency on the UK Continental Shelf was 76% in 2025. The measure compares what fields produced with the most they could have produced. Between 2023 and 2024 it had fallen by 2 percentage points.
The authority says operators produced 21,000 more barrels of oil equivalent a day in 2025. Among other causes, it lists more efficient running of facilities and fewer unplanned shutdowns. Over the year the increase came to more than 7.5 million barrels of oil equivalent.
What AI is doing about it today
Equinor reports that AI contributed US$130 million in value creation and savings for the company and its partners in 2025. One use is predictive maintenance: the company monitors over 700 machines with moving parts through 24,000 sensors across all its facilities to predict failures and maintenance needs.
Equinor says predictive maintenance alone has created US$120 million in value since 2020. It says the work improves safety and reduces the risk of sudden shutdowns that can lead to flaring. These are the company’s own figures, and its release does not say how they were measured.
What we would do
What you would have
You would have an account of what caused recent shutdowns and a ranked list of the equipment that matters most to production. You would also have a documented comparison of monitoring tools, alerts your maintenance planners act on and a monthly report of production lost.
The problem
Canada’s amended methane regulations, published in December 2025, apply to onshore upstream oil and gas facilities. From 1 January 2028 every facility must inspect for leaks on a set schedule. A leak of 100 kilograms an hour or more must be dealt with within 24 hours, and one of 10 to 100 kilograms an hour within seven days.
The government estimates that this leak detection and repair program will cost industry C$3 billion in present value from 2028 to 2040, across about 447,800 sites. It names the time taken to conduct a survey as the main driver of the cost of each survey.
What AI is doing about it today
The UN Environment Programme (UNEP) reports that it uses AI on data from more than 30 satellite instruments. The AI distinguishes methane emissions, such as leaks from oil and gas facilities, from background noise. Its analysts can process 12 to 15 times more data, and an analyst checks every detection the AI flags before anyone is notified.
UNEP says its AI-assisted work identified 80–85% of confirmed detections before expert review, and that it helped mitigate sources estimated to have emitted 1.2 million tonnes of methane.
The government’s estimates, discounted at 2% a year to 2025. The labels are the regulator’s own.
What we would do
What you would have
You would have a measured account of how long leaks take to repair at each site and a documented choice between the two routes the regulations allow. You would also have one record of readings for each facility, automations that turn readings into repair orders and a leak record ready for the regulator.
The problem
The US Energy Information Administration reports that chemicals accounted for 41% of the energy US manufacturers consumed in 2022, and petroleum and coal products for 25%. The figures include purchased electricity and energy sources used as raw materials, so they are wider than the fuel and steam a plant burns. No other kind of manufacturer accounted for more than 11%.
What AI is doing about it today
ENEOS Materials reports a field test in which AI controlled a distillation column at one of its chemical plants for almost a year. Until then, the valves on such a column had been set by hand on the judgment of experienced staff. On that column the AI reduced steam consumption and carbon dioxide emissions by 40% against conventional manual control, measured against the steam previously used to maintain the liquid level. The product met shipment standards throughout, and in March 2023 the company announced that it would adopt the system.
ENEOS reports that it started continuous autonomous operation of a crude oil distillation unit at one of its refineries, run by AI, in January 2024. The system watches 24 key operating factors and adjusts 13 valves at the same time. The company says it has shown higher stability and efficiency than manual operation, and its release gives no figures.
From the agency’s 2022 survey of energy use in manufacturing, published in March 2026. Petroleum and coal products is the survey’s own grouping and is wider than refineries alone. The agency notes that rounded shares may not add to 100%.
What we would do
What you would have
You would have a measured account of where each unit’s energy goes and a ranked list of the units where steadier running would matter most. You would also have a documented comparison of control tools, a safety and approval record for each trial and a monthly report of energy use by unit.
The problem
The World Steel Association reports that making a tonne of steel took 20.95 gigajoules of energy in 2024, against about 50 in the 1960s. It calls energy a significant portion of the cost of steel production. It lists the technology used to control operations among the things a facility’s energy efficiency depends on.
In the US Energy Information Administration’s figures, primary metals producers accounted for 7% of the energy US manufacturers consumed in 2022.
What AI is doing about it today
BHP reports that AI-powered plant control at its Escondida copper mine has saved 3 billion litres of water and 118 gigawatt-hours of energy since its 2022 financial year. The company’s release, of May 2025, does not say how the savings were measured.
Outokumpu, a stainless steel producer, reports that AI lets it optimize melting time in its electric arc furnaces. It says melting time is 4% shorter and furnace capacity 17% higher, which it puts at 1,736 tons less carbon dioxide. Its article, of November 2023, gives no period for these figures.
What we would do
What you would have
You would have one view of the energy used for each tonne by line, crew and product, and a ranked list of where it varies most. You would also have tools that advise your operators and people trained to question them and to work with them.
The problem
A grid operator pays to keep supply and demand matched through every hour, and customers carry the cost. Great Britain’s National Energy System Operator (NESO) reports that balancing the electricity system cost £2.7 billion in 2024/25, 10% more than the year before. Balancing charges made up about 3.4% of an average household’s electricity bill. NESO expects the costs to peak at about £8 billion in 2030, a peak it says faster network building could avoid.
Forecasts bear on that bill, and they are getting harder to make. In its March 2026 forecasting strategy, NESO says the accuracy of its forecasts within the day affects balancing costs and security of supply. It expects installed wind and solar capacity to double by 2030 and forecast errors to grow roughly in proportion.
What AI is doing about it today
NESO has worked with Open Climate Fix on a solar forecast built with machine learning. The first version went into its control room in December 2022. In July 2024 NESO reported that the new national forecast was 2.8 times better than its previous one, for forecasts up to two hours ahead.
NESO’s innovation annex puts a value on the work. It gives the project’s result as a 20% reduction in average forecast error, which is measured on a different basis from the two-hour comparison of July 2024. By using the project’s forecasts in its operations, NESO expects to reduce forecast error by about 50 megawatts. At a wholesale price of £80 a megawatt hour, it estimates this could save £35 million in one year and £175 million over five.
Reported by NESO in July 2024. The lower figure is the best national forecast the project had developed by then; the higher is the forecast NESO was using before.
National Energy System Operator (NESO), Great Britain, 2024.
What we would do
What you would have
You would have a measured account of your forecast errors and what they cost, and a documented comparison of the tools open to you. You would also have a tool tested beside your current forecast, trained forecasters and a monthly accuracy report for your executives and regulator.
The problem
The US Energy Information Administration reports that wind and utility-scale solar generated a record 17% of US electricity in 2025, against less than 1% in 2005. Output that depends on the weather has to be forecast before it can be sold.
Wholesale electricity prices can also fall below zero in some hours. The IEA reports that in South Australia and California, wholesale prices were negative in around 20% of market hours in the first half of 2026. It links negative prices to a lack of flexibility in the power system.
What AI is doing about it today
Great Britain’s National Energy System Operator began a solar forecasting project in 2021. It applied AI to weather data, satellite images and actual solar output, and the operator reports that it cut average forecast error by 20%. In July 2024 it reported that the new forecast was 2.8 times better than its previous one for forecasts up to two hours ahead. The two figures are measured on different bases.
By using the new forecasts in running the system, the operator expects to reduce forecast error by about 50 megawatts. At £80 a megawatt-hour, it estimates the saving at £35 million in one year and £175 million over five. The savings are the operator’s estimates and have not been measured.
What we would do
What you would have
You would have a measured account of forecast error and of the value lost in negative-price hours at each site, and a documented comparison of forecasting options. You would also have automations your teams use each day, people trained to question a forecast and a monthly report for your leaders.
The results described here were published by the organizations named. These are examples of how we would approach the work, and none describes an engagement of our own.
We read each of these at its source. The summaries are ours.
Analysis on planned use of artificial intelligence by businesses in Canada, third quarter of 2026
Statistics Canada
Planned AI use by industry in Canada, with mining, oil and gas and manufacturing below the all-industry figure.
Energy and AI: AI for energy optimisation and innovation
International Energy Agency
The energy agency lists the uses of AI in oil and gas and the barriers to wider use in energy.
Canada’s Energy Future 2026: Executive summary
Canada Energy Regulator
Canada’s energy regulator sets out four scenarios for Canadian natural gas and crude oil production to 2050.
Primary Mitigations to Reduce Cyber Threats to Operational Technology
CISA, FBI, EPA and US Department of Energy
Five measures for owners of control systems, including secured remote access and tested manual operation.
CSB Releases Volume 4 of Incident Reports
US Chemical Safety and Hazard Investigation Board
Serious 2025 accidents, including a refinery fire, linked to weak safety management of how equipment is opened.
Canadian Mining Labour Market Outlook 2026
Mining Industry Human Resources Council (MiHR)
The mining labour council forecasts how many hires Canadian mining needs to 2035 and how many entrants it expects.
Lawrence Berkeley National Laboratory (US Department of Energy), via OSTI
The capacity waiting in US grid connection queues and how long connection takes.
International Energy Agency
The energy agency compares how long grids and renewable projects take to complete and counts the projects stalled in queues.
2025 Long-Term Reliability Assessment
North American Electric Reliability Corporation (NERC)
The North American reliability organization forecasts how fast peak demand for electricity will grow and which regions may fall short of supply.
Evidence last reviewed on 6 October 2026.
A briefing is a confidential conversation with the executive accountable for a technology decision. We set out how we would approach the question and what evidence it would require. Scope and fees are agreed in writing before any work begins.
