AI is reshaping what data centers need from infrastructure systems. For example, access to power and grid capacity can shape where – and how quickly – AI infrastructure scales. Part of the solution is a broader mix of energy sources alongside more efficient, resilient infrastructure.
According to the International Energy Agency’s latest outlook, global data center electricity consumption is projected to roughly double, from 485 TWh in 2025 to approximately 950 TWh by 2030. For customers and partners across the data center ecosystem, the challenge is not only securing enough electricity but having it available around the clock—with the reliability and affordability their operations require, and an emissions profile aligned with their goals.
CorPower Ocean, a Cisco Investments portfolio company, has explored how wave energy technology could complement other sources for 24/7 energy needs. A new CorPower white paper brings real-world data center load characteristics into the conversation.
Why the AI era needs a broader energy mix #
A data center is a flat-load customer: demand does not disappear at night or when weather conditions change. That means adding lower-carbon megawatt-hours but also building a portfolio that can match demand hour-by-hour while limiting volatility, curtailment, and storage needs.
Different resources bring different strengths. Solar can provide abundant daytime energy. Wind can contribute across day and night but varies with conditions. Batteries can shift energy across hours. Wave energy can add another useful generation profile because waves often persist overnight and across seasons, at times when solar and wind output may be lower.
Cisco is committed to exploring emerging energy technologies and other approaches that can help customers across the data center ecosystem meet their 24/7 energy requirements. That work also includes making AI-ready infrastructure more efficient and giving operators better visibility and control over energy use – priorities reflected in Cisco’s Plan for Possible.
What CorPower’s new white paper shows #
In a new white paper, CorPower modeled a hypothetical 100 MW flat-load data center in the San Jose, California, area. The model used anonymized operational 2024 electricity records from a commercial hyperscale data center to shape a realistic, always-on load profile.
The study compared cost-optimized energy portfolios with and without wave energy, drawing from wave, solar, wind, batteries, and limited grid purchases. The model sought to supply at least 95% carbon-free electricity while minimizing the total system cost.
The modeled 2032 scenario shows a portfolio with wave energy compared with one without it across four system-level metrics relevant to always-on loads:
- A smaller overall system: installed generation capacity falls from628 MW without wave to386 MW with wave – about 39% less capacity.
- Far less storage needs: modeled energy storage capacity falls from3 GWh to14.6 GWh – more than 90%.
- Less wasted energy: energy sold or curtailed falls from49% to37% , indicating a better match between generation and demand.
- Lower cost of delivered electricity: levelized cost of electricity falls from**$114.5/MWh** to**$84.6/MWh** – a modeled reduction of about 26%.
CorPower also modeled 2040 and 2050 planning horizons. In those scenarios, including wave energy reduced modeled system cost by 39% in 2040 and 45% in 2050. These are scenario results, not measured deployment outcomes, but they illustrate the potential value of a resource that generates on a different schedule from solar and wind.
From exploration to a pathway for scale #
- Proven at sea: CorPower’sC4 wave energy converter has exported electricity to the grid from the Aguçadoura site off northern Portugal and operated through severe Atlantic conditions.
- Independent validation: in July 2026, CorPower announced that the C4 received theworld’s first DNV Prototype Certificate for a wave energy converter.
- A path toward industrial scale: CorPower is advancing its first industrial wave farms in Portugal and Scotland, building on a modular approach intended to scale from individual devices to larger arrays.
- A globally relevant use case: early deployments are in Europe, but this white paper applies the technology to a California data center scenario. That shift shows the benefits of evaluating wave technology wherever suitable wave resources and always-on energy demand come together.
Powering the AI era takes more than generation #
The energy challenge for AI infrastructure is bigger than generation alone. New sources need to be connected to the grid, electricity must move reliably from generation through distribution, and operators need secure connectivity, telemetry, and automation to manage complex systems. In other words, the future of energy depends on generation, distribution, and networking working together.
Cisco engages across these layers. Cisco’s utility network solutions support secure, scalable connectivity from power generation to distribution and metering. And Cisco’s work on intelligent data center energy management gives operators visibility, insights, and automation to manage energy use while protecting availability and performance — alongside our ongoing work on energy distribution and energy networking (learn more about our work on power).
We are also applying these priorities in our own operations. In fiscal 2025, Cisco sourced renewable electricity to match 100% of global annual electricity needs at Cisco-owned and leased facilities. The next frontier includes pairing efficiency with a broader set of energy sources and technologies.
Wave energy will not be the only answer, and no single technology will meet every need. But CorPower’s progress and the whitepaper’s findings make a strong case for including wave in the conversation. Cisco will keep exploring innovations across generation, distribution, and networking that can help hyperscalers and data center operators meet 24/7 energy requirements while building a more resilient energy system and more sustainable AI infrastructure.
Transparency note*: The whitepaper models a hypothetical 100 MW data center near San Jose, California, using an anonymized 2024 load profile from a commercial hyperscale data center. Results are modeled scenarios, not measured outcomes from a Cisco data center or an operating wave-powered data center. The views in the paper are those of its authors and do not necessarily reflect the views of Cisco.*
Power an Inclusive Future for All
Driven by our Purpose, we combine our technology, people, and broader networks to address society’s greatest challenges.