For decades, the North American energy model was predicated on generating power in massive facilities and pushing it down a one-way street to customers and businesses. That era is now coming to an end.

The retirement of coal burning facilities has been underway for some time. As the US and Canada continue to work to integrate green energy and new nuclear resources into the grid to backfill that capacity, we are also now witnessing a surge in demand to support the AI industry and infrastructure for electric vehicles, which in turn is driving concerns at a local and national level around environmental impacts and integrity of supply.

Highlighting the potential future capacity crunch, regional transmission organization PJM Interconnection, which serves 13 states and the District of Columbia, expects the net energy load to grow by an average of 4.8% annually for the next decade – a wholly unprecedented rise.¹

There’s more to addressing this emerging scenario than building extra capacity ‘to keep the lights on’. The challenge for North American utilities, from the reliability reforms underway in ERCOT in Texas to evolving markets in Canada, is to build a fluid system where price signals, resource availability, and congestion are managed in real time so that energy can be directed to where it is needed at the most affordable price.

The money is there, with more than $700 billion in investments targeted for grids and generation sources between now and 2030.² So is the will. Energy generators, distributors, governments and regulators are now all focused on the construction of a unified grid where operational synergies will determine grid output and the cost of energy production and delivery will ultimately drive monthly consumer bills.

We have identified the five key trends that will shape the direction of the North American energy sector as it rewires for the future.

1. From monolith to flex-grid: the end of static capacity

Historically, utilities relied on long forward capacity lead times from stable assets such as coal. Today, legacy generation facilities are being retired due to multiple factors, leaving a gap that is not easily filled. Certainly, solar, wind and small nuclear (SMRs) all have a role to play, but a more flexible, unified grid which supports dynamic load management is also required.

In 2026, we will begin to see more pilots and programs that demonstrate the viability of such systems, capable of sending simultaneous price and operational signals for next generation demand response and power optimization. These models coordinate resources based on real-time grid congestion and availability. Also known as load flattening, they will help meet growth in demand driven by AI and electric vehicles.

For North American operators, this means the distinction between generation, load and grid is beginning to blur. When a utility (whether deregulated or regulated) is no longer bound to inflexible legacy operating models, systems can be modernized in line with so-called ‘power optics’ – the blend of price, demand, and generation signals – that govern the grid’s ability to flex rather than break under pressure.

2. The $5 trillion wave: segmenting data center demand

In 2026, the influence of the technology sector on the grid is no longer marginal, but dominant. With an estimated $5 trillion of capital investment sitting upstream of data center capacity, the pressure to bring these resources online will force operators to bend operational models.³

However, 2026 will see the industry stop treating all data centers as equal and categorize them according to load shapes. Utilities must distinguish between:

• Hyperscalers – operations with fluctuating loads depending on training activity
• Crypto mining – highly intermittent and price-sensitive loads
• Colocation – the rise of third-party data centres that have an ‘always on’ service model and require flat, constant power.

Each of the above have distinct and unique energy consumption profiles and needs. The unified grid approach must treat these large loads almost as intermittent resources themselves, with each asset class having unique operational considerations. To meet operational demand, we expect to see additional data center facilities built with a ‘behind-the-meter’ component i.e. data centers that col-locate with energy generation facilities or integrate their own generation capacity, such as a co-located midstream gas facility, which is not yet integrated with the grid. The challenge for next year is to bring that private capacity into the public grid to balance wholesale and retail price signals.

3. Agentic AI and the affordability imperative

2026 is the year where AI comes into its own in helping to power the grid. With consumer electricity bills in some regions rising by nearly 30% in recent years – driven by factors including inflation, tariffs and capacity challenges – optimizing operations and reducing costs is a priority.⁴ The industry cannot pass all implementation costs down to the residential ratepayer.

One prospective solution lies in scaling existing AI tools and advancing agentic AI solutions that automate routine tasks and analytics and drive operational efficiency across the value chain. We anticipate the adoption of AI-enhanced systems to improve weather forecasting, optimize load dispatch, and support localized grid optimization. 

With rising customer expectations, the shift from channel consistency to channel continuity is essential for customer satisfaction and loyalty. Providers must ensure seamless customer context across all platforms, from chatbots to phone calls. This includes the design and deployment of emotionally intelligent self-service solutions that offer empathy and advice to all customers, including vulnerable populations.

Furthermore, we expect a renewed focus on the Advanced Metering Infrastructure (AMI) already installed across several US regions including the west coast and northeast.  Currently, this data is underutilized. By synthesizing smart meter data, utilities can protect residential customers from price spikes, while enabling sophisticated ‘Virtual Power Plant’ (VPP) models, which turn flexible customer assets into marketable grid resources, for those who wish to participate.

4. Market dynamics: the rise of the ‘prosumer’ giant

The commercial relationship between big tech and big energy is being rewritten. Companies such as Amazon, Meta and Google are no longer satisfied with being passive customers and are seeking to directly participate in the grid. In 2026, we accordingly predict the growing adoption of Hybrid Peak Power Purchase Agreements (PPAs).

In this new model, the behind-the-meter asset (such as a renewable installation or gas backup) shares the responsibility for load management with the grid operator. This shifts the risk profile and requires new regulatory frameworks. We are already seeing friction and adaptation in specific jurisdictions:

• Ohio – new tariffs are requiring large loads in the queue to pay at 85% capacity regardless of usage, an attempt to pass the cost burden onto the new entrants rather than the ratepayer⁵
• Texas (ERCOT) – the market is already innovating with residential participation, but other North American regions must catch up to allow similar flexibility.

For Canada and the broader US, adapting engagement models to accommodate these non-traditional players will be the defining commercial challenge of the year.

5. Holistic resilience: cyber, storms, and demand risk

The convergence of Information Technology (IT) and Operational Technology (OT) requires a more resilient approach to the grid as it becomes more digital, and vulnerable. Advanced intrusion detection and the ability to segment and isolate at-risk areas of the grid to prevent the cascading failures seen during 2021’s Winter Storm Uri are now non-negotiable.

In this scenario, cybersecurity emerges from the back-office and becomes a key component of customer trust and operational resilience. As energy providers transition to digital models by combining real-time grid data with customer information, the traditional separation between corporate IT and grid control systems is fading.

This change requires coordinated security models that balance IT confidentiality with availability and safety. Legacy vulnerabilities remain a challenge, as many operational components run on outdated hardware and lack modern security measures, making them vulnerable to cyber threats as they become more interconnected.

To mitigate these risks, many providers are implementing a Zero Trust approach, which involves continuous validation of every device and user across all systems and endpoints. There is also a focus on cyber-physical resilience, enabling manual overrides and fail-safe operational mechanisms that maintain grid stability during cyber-physical attacks.

Demonstrating maturity in this converged environment is now crucial for regulatory approval, as commissions require detailed cyber risk reports for rate-setting. But by embedding security by design, organizations can transform existential cyber risks into a competitive advantage, enhancing long-term customer and regulatory confidence.

In 2026 we also anticipate the emergence of demand risk. Right now, the industry is rushing to expand infrastructure for AI hyperscalers. But what happens when a new, highly efficient chip model is released, depleting forecasted energy demand overnight?

If demand evaporates after the infrastructure is built, who pays for the stranded assets? True resilience requires protection against such financial volatility. We must ensure that long-term infrastructure planning accounts for the possibility that the current AI energy hunger is a bubble, or at least subject to rapid efficiency gains.

Conclusion: orchestrating the future