Gas, oil and electricity: what really powers the next investment boom?
Higher gas and oil prices make electrification more attractive. They do not automatically make electricity cheap, or every clean-energy project profitable. Gas can raise the wholesale power price; oil changes the cost of driving; and grids determine whether new generation can actually reach a car charger or an AI data centre. These are connected problems with different solutions.
For Europe, our central judgement is that gas-price volatility will remain a planning risk through the next two winters, while additional supply could ease prices later this decade. For AI, access to dependable power at a particular site is a more convincing constraint than a worldwide shortage of electricity. For passenger cars, affordable vehicles and convenient charging often matter more than national generation capacity.
Analysis dated 29 September 2026. Focus: Europe and France, with US comparisons. Observations, institutional forecasts and our own conditional scenarios are distinguished below. Numerical examples are illustrative assumptions, not current tariffs or price quotations.
Why gas can set the electricity price
A power system must match supply and demand continuously. In a simplified electricity auction, available offers are accepted from cheapest to most expensive until demand is met: the merit order. The last accepted offer sets the clearing price for that delivery interval and market area.
This is not the highest price anyone has ever paid, nor the cost of the most expensive plant in Europe. An expensive plant whose offer is not needed does not set that auction’s price. The European Commission describes this pay-as-clear system and explains why replacing it with pay-as-bid would also change producers’ bidding behaviour, rather than simply guarantee cheaper electricity. European Commission: electricity market design.
Consider a deliberately simplified hour:
| Available supply | Offer price | Cumulative capacity |
|---|---|---|
| Wind and solar: 40 GW | €0/MWh | 40 GW |
| Nuclear and other low-variable-cost generation: 35 GW | €20/MWh | 75 GW |
| Gas: 25 GW | €100/MWh | 100 GW |
At 90 GW of demand, 15 GW of gas is needed and the clearing price is €100/MWh. At 70 GW, the €20/MWh block is sufficient. If demand stays at 90 GW but gas offers rise to €140/MWh, the price becomes €140/MWh. These are teaching assumptions, not representative offers or a dispatch forecast; real markets include network limits, imports, storage and complex bids.
That explains how gas can produce a minority of electricity yet influence its price in many intervals. It also explains why adding solar can lower the midday price without eliminating exposure to gas on a winter evening. Interconnection carries price effects across borders, but congestion can separate neighbouring market prices.
A €10 gas increase can mean roughly €18 more for gas-generated electricity
For a gas plant, a simplified variable-cost calculation is:
Electricity cost = gas price ÷ efficiency + carbon cost per electrical MWh + variable operating cost.
Assume 55% efficiency, emissions of 0.36 tonnes of CO₂ per electrical MWh, a carbon allowance price of €70/tonne and €4/MWh in other variable costs. At gas costing €30 per thermal MWh, the total is approximately €84/MWh of electricity. At €60 gas, it is approximately €138/MWh.
The difference is €30 ÷ 0.55 = €54.55/MWh, or 5.45 euro cents/kWh. A €10 increase in gas alone adds €18.18/MWh. These are our calculations with fixed assumptions; start-up costs, heat sales, plant constraints and changing carbon prices are excluded. The underlying conversion matters: fuel energy bought is larger than electrical energy sold. ACER–CEER: retail energy market report, explanation of gas-price transmission.
This is a sensitivity for a gas plant, not a prediction that every electricity bill rises by 5.45 cents. Gas must be setting the relevant market price, and the supplier must actually be exposed to that price.
A regulated bill is not a permanently capped commodity price
Three different things are often called an electricity “cap”: a regulated retail tariff, temporary government protection for consumers, and a wholesale trading limit. None makes fuel, network maintenance or new infrastructure free. If government shields consumers from a cost increase, someone still carries the difference through subsidies, supplier obligations or subsequent adjustments.
In France, the regulated tariff combines electricity procurement, network charges and taxes. CRE’s January 2026 explanation says the market procurement component is smoothed over two years. That averaging can delay both rises and falls: today’s bill does not simply reproduce today’s spot market. CRE: construction of the regulated tariff, January 2026.
For a household or factory, ask what the contract fixes: the energy component, the full unit price, the subscription, or only a period before renewal. For a utility, distinguish its businesses. An unhedged generator, a fixed-price retailer buying wholesale power, and a regulated network operator have very different exposures. A high power price can benefit the first, hurt the second and say little about the third’s allowed return.
Long-term power purchase agreements (PPAs) and two-way contracts for difference can reduce exposure to short-term prices. In a simplified two-way contract, the producer receives support when the reference price is below the agreed level and pays back when it is above. The details of volume, reference price and settlement still matter. European Commission: questions and answers on electricity market reform.
Which investments benefit when gas becomes expensive?
The useful question is not “which technology wins?” but which cash flow changes, for how long, and who bears the risk? The following is our economic interpretation of those exposures.
| Investment | Effect of higher gas prices | What can prevent a profitable project? |
|---|---|---|
| Solar and wind | Avoided gas generation becomes more valuable; unhedged output can earn more in gas-priced intervals | Low prices when output is abundant, curtailment, connection delays and financing costs |
| Batteries | Larger differences between cheap and expensive intervals can improve trading opportunities | High average prices alone do not guarantee a useful spread; losses, degradation and competing batteries matter |
| Existing nuclear | Available output becomes more valuable when it displaces expensive fossil generation | Outages, maintenance, safety requirements and contracts limiting price exposure |
| New nuclear | A durable expectation of expensive fossil power can strengthen the long-term case | Construction time, overruns, financing and future revenue certainty |
| Grids | Connecting cheaper supply and flexible demand can save more system-wide | Permits, equipment, cost allocation and regulatory approval |
| Electric cars | Indirect benefit only if the electricity price remains favourable relative to petrol or diesel | High charging prices can erode savings even when oil is expensive |
Solar: the price at noon matters more than the annual average
A solar farm earns the price available when it generates. If many farms produce together, their output can depress those same prices. Its generation-weighted selling price—its “capture price”—can therefore fall even while evening electricity remains expensive. The IEA documents the growing challenge of negative prices and renewable integration. IEA: Renewables 2025, renewable electricity.
For a rooftop system, value depends on self-consumption, the tariff actually avoided and payment for exported power. An illustrative 3,000 kWh consumed on site saves €750 at an avoided variable rate of €0.25/kWh, versus €540 at €0.18/kWh. The €210 difference can materially change payback; neither calculation eliminates fixed subscription costs. A battery may increase self-consumption, but must earn back its own cost and losses.
Nuclear: distinguish electricity already available from a promise to deliver it
An existing reactor, a proposed lifetime extension and a new reactor are three different investments. New nuclear may provide valuable dependable low-carbon generation, but a temporary gas spike cannot resolve its construction and financing risks. The IEA’s nuclear investment study identifies financing, project delivery, workforce and supply chains as central challenges. IEA: The Path to a New Era for Nuclear Energy.
For an AI campus seeking power in the next few years, a future reactor helps only if its commissioning date matches the campus’s needs. Contracting existing output can secure commercial supply for a buyer without immediately adding generation to the system. New projects, restarts, uprates and life extensions must each be assessed on their own delivery assumptions.
Grids: higher social value does not automatically mean higher investor returns
A new line can connect low-cost production to an expensive region. A stronger distribution network can serve a charging depot. But the resulting savings are spread among users, so a developer cannot necessarily collect the full benefit through market prices. Regulated cost recovery, connection agreements and who pays for upgrades are decisive.
The IEA’s Electricity 2026 identifies more than 2,500 GW of generation, storage and large-load projects in connection queues. It estimates annual grid investment needs to rise roughly 50% from about $400 billion to meet demand through 2030. Queued capacity is not a forecast that all those projects will be built. IEA: Electricity 2026, grids.
Expensive energy can also make construction more expensive. A project benefits only if improved expected revenues or avoided costs outweigh higher equipment, labour and financing costs. An energy shock that raises interest rates can particularly hurt investments whose costs arrive upfront and whose savings arrive over decades.
Oil changes the equation for transport more directly
Oil and natural gas are different commodities. Petrol and diesel are made from oil; gas is an important input to electricity generation. They can rise together after a shared supply shock, but not in a fixed ratio. Gas remains constrained by pipelines, liquefaction, shipping and import terminals; oil-linked LNG contracts create another channel of connection. US EIA: why gas markets differ from oil markets.
Our interpretation: higher oil prices most directly encourage efficient vehicles, EVs, rail and reduced driving. The incentive for solar, nuclear or electricity networks is more indirect, through electrification, some fuel-contract linkages and broader energy-security policy. In systems dependent on oil-fired generation, the direct electricity effect is larger.
Nor does a 20% rise in crude guarantee a 20% rise at the pump. Refining margins, taxation, currency movements and distribution intervene. A joint oil-and-gas shock can make both driving and charging dearer; what matters to an EV buyer is their relative cost.
A worked EV comparison
Assume a petrol car uses 6 litres/100 km and an EV purchases 18 kWh/100 km at the charger, including charging losses. Hold vehicle efficiency constant. These are scenarios, not national averages.
| Scenario | Petrol price | Electricity price | Petrol cost/100 km | EV cost/100 km | EV energy saving |
|---|---|---|---|---|---|
| Starting example | €1.80/litre | €0.25/kWh | €10.80 | €4.50 | €6.30 |
| Oil rises; charging unchanged | €2.20/litre | €0.25/kWh | €13.20 | €4.50 | €8.70 |
| Oil and charging both rise | €2.20/litre | €0.40/kWh | €13.20 | €7.20 | €6.00 |
| Expensive public charging | €2.20/litre | €0.75/kWh | €13.20 | €13.50 | −€0.30 |
The energy-cost break-even charging price is litres/100 km × petrol price ÷ kWh/100 km. With these assumptions it is €0.60/kWh at €1.80 petrol and approximately €0.73/kWh at €2.20 petrol. Session, parking or additional subscription fees would need adding.
At 15,000 km/year, the starting example saves €945 in energy; the oil-only increase raises that to €1,305. An assumed €5,000 additional purchase cost would take about 5.3 or 3.8 years respectively to recover from energy savings alone. This deliberately excludes financing, depreciation, insurance, maintenance, incentives and charger installation: fuel savings are not total ownership cost. The IEA likewise distinguishes charging access and charging prices when examining EV economics. IEA: Global EV Outlook 2026, electric vehicle charging.
Is electricity really the transport bottleneck?
Separate energy from power. Energy is the quantity consumed over time, measured in kWh or TWh. Power is the rate at one instant, measured in kW or GW. One GW running continuously for a year uses 8.76 TWh.
An illustrative one million EVs, each driving 12,000 km/year and purchasing 18 kWh/100 km, require 2.16 TWh/year. That averages only about 247 MW. Yet if all one million charged simultaneously at 7 kW, they would demand 7 GW. This extreme comparison is not a forecast; it shows why charging schedules and local equipment can matter more than the annual total.
Charging spread through available hours can use spare capacity. A motorway charging hub or truck depot creates a much more concentrated connection requirement. Some drivers also lack a private parking space, so cheap domestic electricity is not an available option. For heavy trucks, aviation and shipping, vehicle duty cycles, payload and energy-storage requirements add constraints that a passenger-car calculation cannot resolve.
France is a useful counterexample to an indiscriminate electricity-shortage story. RTE reports 547.5 TWh of generation in 2025, more than 95% low-carbon, and record net exports of 92.3 TWh. This supports an opportunity to electrify, not a claim that every substation has spare capacity or that power is guaranteed through every winter peak. Exports also serve neighbouring systems. RTE: Bilan électrique 2025.
Why AI makes the local constraint harder
The IEA’s updated 2026 projection puts all data-centre electricity consumption at about 950 TWh in 2030, up from 485 TWh in 2025 and around 3% of global electricity demand by 2030. This includes non-AI computing; it must not be presented as AI’s consumption alone. The agency also identifies constraints in chips, capital and energy equipment, so electricity is not the only possible limit. IEA: Key Questions on Energy and AI.
A large campus concentrates demand in one place. As a transparent example, a 1 GW average facility load, including cooling and other overhead, needs 8.76 TWh/year. If 1 GW refers only to IT equipment and assumed power usage effectiveness is 1.2, facility demand becomes 1.2 GW and 10.51 TWh/year. Announced connection capacity, actual average draw and server capacity are not interchangeable.
Buying enough solar generation over a year does not ensure matching supply every night. A dependable supply plan needs generation, network access, balancing and backup that work together. For scale, a 1 GW/4 GWh battery supplies a 1 GW load for at most four hours before losses and operating reserves; it is not by itself a solution to a multi-day supply gap.
In our assessment, the competitive advantage is therefore a credible date for energisation, a deliverable connection and a dependable supply contract. Some training work can move in time or location; latency-sensitive services have less freedom. Operators must establish how much flexibility they can actually offer, rather than assume all computing can stop whenever power is scarce.
Onsite gas generation can bypass some grid delays, but it adds fuel exposure and still requires equipment, permits and reliable gas delivery. The IEA cautions that turbine supply constraints can undermine its presumed speed advantage. IEA: 2026 assessment of onsite power for data centres.
AI need not raise everyone else’s bill. The outcome depends on spare capacity, new investment and who pays for it. Our policy judgement is that large users should provide credible demand commitments and contribute to the incremental infrastructure they require; speculative connections should not leave households paying for unused assets.
Will gas stay high? A conditional outlook, not a single price target
First specify where, compared with when, and for which delivery period. European TTF, Asian LNG and US Henry Hub are not one price. Here “persistent high prices” means sustained pressure relative to Europe’s 2025 conditions, not a claim that every region returns to the 2022 extremes.
The IEA’s July 2026 report describes a major disruption to Gulf LNG supply and delayed relief from new capacity. It reports a second-quarter TTF average near $16/MMBtu, 32% above a year earlier. That is a historical quarterly average, not a 29 September quote. Its forecast assumed a third-quarter reopening of Hormuz: that assumption must not be mistaken for an observed outcome. IEA: Gas Market Report, Q3-2026.
A more recent cross-check is the EIA’s 9 September 2026 outlook, whose model inputs closed on 3 September. It projects Brent around $90/barrel in the second half of 2026, easing to a $74 annual average in 2027, conditional on recovering supply. Its US Henry Hub annual forecasts are $3.43/MMBtu in 2026 and $3.28 in 2027. These are forecasts with different regional and time bases, not directly comparable spot quotes and not a European gas prediction. EIA: September 2026 Short-Term Energy Outlook, pages 2–3.
Our scenarios below are qualitative analytical judgements, not IEA scenarios or statistically calibrated probabilities.
| Scenario | Conditions | Implications for 2026–2028 | Implications toward 2030 |
|---|---|---|---|
| Central: gradual, uneven normalisation | Supply recovers, new LNG arrives, no sustained new interruption | European gas remains vulnerable to winter spikes; no reliable straight-line fall | Additional supply and electrification can reduce pressure, while local power-network constraints persist |
| Persistent supply stress | Repairs or shipping recovery disappoint; cold weather or weak hydro adds demand | Gas and gas-sensitive power stay expensive; fuel substitution and demand destruction increase | Stronger incentive to diversify, but financing and construction costs can delay the response |
| Faster easing | Mild weather, weak demand, successful supply recovery and strong low-carbon output | Gas prices fall more quickly; fewer expensive gas-setting intervals | Electrification still has an efficiency case, but merchant generation must survive lower selling prices |
Our forecast is more confident about recurring volatility and local connection scarcity than about a permanently high gas-price level. For Europe, projects should remain viable through a stressed winter and through a later period of cheaper gas. For the United States, do not import the European LNG-price story wholesale. For 2028–2030, easing is plausible, but dependent on delivered supply and demand—not announced capacity alone.
To revise that judgement, watch actual LNG shipments and commissioning, storage relative to seasonal norms, winter weather, Asian LNG demand, nuclear and hydro availability, and TTF prices across delivery years. Futures prices are useful market signals, not guarantees. Natural-gas supply, storage, trade and weather all influence the balance. EIA: factors affecting natural-gas prices.
What a resilient investment decision looks like
For a solar or nuclear project, test realised output prices and financing costs under both high- and low-gas scenarios. For storage, test the spread between charging and selling prices after losses. For an EV, use the charging options actually available to the driver. For a grid upgrade, establish what it enables, when, and how its cost is recovered.
New wires are only part of the response. The IEA’s September 2026 grid-modernisation study highlights better use of existing networks through dynamic ratings, network reconfiguration and power-flow control. Their value is site-specific; they complement reinforcement rather than remove every physical limit. IEA: Modernising Grids in the Age of Electricity.
For an AI project, scrutinise the promised power date as closely as the promised computing performance. A generation contract, an equipment order and a completed grid connection are different milestones. Efficient computation lowers energy per task, but the total also depends on how many tasks people buy and how demanding they become.
The strongest conclusion is therefore practical: high fossil-fuel prices increase the value of electrification, but dependable delivery determines its speed. The next successful investments will need affordable generation, workable financing, flexible demand and connections that arrive on time. A bet on permanently expensive gas is a weaker foundation than a project that remains useful when gas becomes cheaper.
Method: desk research using primary publications from the IEA, EIA, European Commission, ACER, CRE and RTE, consulted on 29 September 2026. Linked sources support reported observations and institutional forecasts. Worked examples and the three scenario paths are our calculations and interpretation; no econometric causal estimate or live commodity-price feed is claimed. Forecasts should be revisited as supply, weather and project delivery change. Sources may update their live pages; dates and reporting periods above identify the editions used.