AI compute infrastructure with renewable energy

🚨 Google just made the grid optional ⁉️ ... Let that sink in! “Build the power first - then the data centre.” That’s Google’s new approach. And it changes everything. In a move that redefines how tech giants approach digital infrastructure, Google is investing $20 billion to build a fleet of AI-optimised data centres. Not just near renewable power sources, but right on top of them. This signals a powerful shift from "data centre-first" to "energy-first" design, accelerating Google's push to deliver AI at scale while staying on track to reach net-zero. 📍 At a Glance Developer: Google Energy Partner: Intersect Power Investment Partner: TPG Rise Climate Scope: Multiple U.S. sites co-locating hyperscale compute with solar, wind, and storage Deployment Timeline: Early phases by 2026, full rollout by 2027 ⚡ Purpose-Built for Performance & Power Infrastructure tailored for AI and high-performance computing Battery-backed energy ecosystems to balance loads and boost uptime Designed to relieve pressure on traditional power grids and accelerate delivery speed 🌿 Sustainability Baked In 100% renewable energy sourced on-site; no offsets, no greenwashing Projected to cut energy transport losses and carbon emissions Aligns with Google’s pledge to run on carbon-free energy, 24/7, by 2030 🌍 Strategic Implications Tackles power availability bottlenecks in hyperscale site planning Encourages broader energy + data co-location strategy for the tech sector Brings major infrastructure investment to underutilised regions rich in solar/wind 🔎 A New Kind of Site Selection Google’s site criteria for this project flips the script: Energy availability, not just fibre and land, is now the lead driver of where these AI data centres are going. We're witnessing a major evolution in how hyperscalers will plan capacity at scale in the 2030s. 💡 Why This Is Bigger Than Just Google This isn’t just about powering AI, it’s about reimagining the global infrastructure it runs on. As cloud providers grapple with rising demand and public scrutiny around emissions, this project sets a powerful precedent for future-ready, responsible growth. 👉 Is energy-first development the next standard for data centres? What do you think: Should more hyperscalers follow suit? #DataCenters #AIInfrastructure #Sustainability #CloudComputing #Google #RenewableEnergy #DigitalInfrastructure #IntersectPower #NetZero #GreenTech #FutureOfCompute

Ever wondered where the future of AI is being built? I just visited the data centre in Finland that's making it happen. Nebius’ data centre is the powerhouse where AI models are trained. Thousands of GPUs working in unison. It’s expanding to host up to 60,000 GPUs dedicated to intensive AI workloads. They’re building a full-stack AI cloud platform. Here’s what I learned: 1. There is a scarcity of GPUs in the US • Clusters are being sold in massive packages • People who need smaller requirements can’t find them 2. Nebius are building a self-serve platform • Cover infrastructure requirements from a single GPU to big GPU clusters • They’re not a GPU reseller—they’re designing the servers and the racks from the ground up 3. Applications • Helped Mistral train their multimodal models • Provide full-stack infrastructure for AI model development Something else that was unique about the visit. Nebius cools the servers in Finland using the outside air. The heat that’s generated from the servers is then shipped back into the grid. This means Nebius not only heats the onsite building, But it also heats homes nearby, benefitting the local community. They’re able to recover 70% of the heat generated. And it’s the first in the world to have this heat reuse application connected to the local municipal grid. They’re now investing over $1B in AI data centres in Europe. I feel the future of AI depends on infrastructure like this that balances performance with sustainability. Follow me Alex Banks for daily AI highlights & insights.

Google just announced a data center in Wilbarger County, Texas, that will be powered by onsite clean energy developed by US energy firm AES. ⚡️ AES has already secured the land and the interconnection agreements. This “power first” approach ensures that the data center and colocated energy infrastructure come online together. Google has signed a 20-year power purchase agreement tied to the colocated assets. 🤝 The structure matters. AES will own and operate the generation, and optimize costs under a long-term energy management agreement. Google brings the compute load and the paying dollars. AES brings powered land, the single scarcest resource in AI infrastructure today. ❗️ Andrés Gluski, AES’s CEO, put a number on it: “To-date, AES has signed agreements for nearly 12 GW of energy with data center customers, 9 GW of these are PPAs directly with hyperscalers.” This is a template that works. Not tech giants trying to become a utility by themselves overnight. The developer-led model where the power comes first and the hyperscaler shows up with a 20-year PPA. 💡 What do you think of this announcement? Share in the comments below. Link to the full article in the comments. ⬇️ Source: Data Center Dynamics 2/25/2026 Disclaimer: Nothing in this post constitutes investment advice. #datacenters #AI #powerdemand #technology #innovation

Saudi Arabia’s renewable-energy strategy isn’t only about clean power it’s about global AI compute leadership. -130 GW (57.5 GW by 2025) clean power is becoming the backbone of low-cost AI compute - A GPU-hour in Saudi can be ≈10–12× cheaper (electricity only) than in a typical global market. - New AI data-center partnerships (NVIDIA, xAI, AWS, AMD, Cisco, Luma, Adobe, Qualcomm, Blackstone/AirTrunk…) are building multi-GW hyperscale clusters on top of this energy advantage. This is what happens when energy policy and AI infrastructure are designed together.

The most interesting part of this project isn't that the data center is underwater. It's how multiple infrastructure systems have been designed to work together. China's new offshore wind-powered underwater data center combines renewable energy, natural seawater cooling, and AI computing in a single integrated system. By using the surrounding ocean for cooling and nearby offshore wind for power, the project reduces both energy demand and freshwater consumption. As demand for AI infrastructure continues to accelerate, the conversation can't stop at computing capacity. We also need to rethink how these facilities are powered, cooled, and connected to the broader energy ecosystem. The future of infrastructure won't be defined by isolated innovations. It will be shaped by systems that create greater value by working together. #DataCenters #SystemsThinking #ArtificialIntelligence #DigitalInfrastructure #EnergyInnovation #FutureReady #Infrastructure #Resilience #BuiltEnvironment Source: https://lnkd.in/gwDDCTEF

Right now, wind farms are being paid to stop producing electricity. It's called curtailment — clean energy generated but stranded because the grid can't absorb it. Billions of dollars of renewable capacity, wasted every year. This is the problem I see up close as a board member at Soluna Computing (SLNH). Soluna's model is deceptively simple: co-locate data centers directly at wind, solar, and hydro sites — turning surplus energy into high-performance computing for AI and Bitcoin workloads. As an investor, I look for companies solving a real physical constraint with a capital-efficient model. Renewable computing does exactly that. The energy transition isn't just about generating more clean power. It's about building smarter infrastructure around the energy we're already wasting. What's the biggest infrastructure inefficiency you see being ignored right now? #EnergyTransition #DigitalInfrastructure #RenewableEnergy #AIComputing #BoardLeadership

The latest power couple? Data centers and renewable energy.  RMI’s latest analysis finds that pairing new data centers with new wind, solar, and energy storage near existing power plants can supply the electricity needed for AI growth without increasing costs for businesses and local communities.  For big tech, having on-site carbon-free electricity for their data centers supplies a reliable energy source, helps avoid transmission upgrade costs, and provides the opportunity to sell surplus energy back to the grid. Even if data center growth slows, the new clean energy resources would still support grid reliability and reduce costs for communities. Many have been asking about the role of AI and data centers in the energy transition, so I encourage you to dig into this work by Alexander Engel, David Posner, and Uday Varadarajan.  https://lnkd.in/gBTA23ty   

More than 4 years — that's how long it takes to build power projects in the U.S., more than double the time it took just 15 years ago. This lag in electricity grid planning, coupled with the urgent need for sustainable solutions, presents a significant challenge to economic growth, especially with AI's potential to add over a trillion dollars annually to U.S. GDP by 2030. To address this, Google is partnering with Intersect Power and TPG Rise Climate to co-locate data centers with new clean energy plants. This innovative approach: 🟢 Synchronizes clean power generation with data center growth, ensuring reliable and carbon-free energy for AI. 🟢 Reduces the timeline to operation by bringing data centers online alongside their dedicated power source. 🟢 Minimizes the need for new transmission infrastructure by building where power is generated. This strategy not only accelerates the transition to a carbon-free future for AI but also helps alleviate grid constraints and improve overall reliability. It's a crucial step towards responsible and sustainable digital infrastructure development. In other words - I think this makes common sense :) Read more here: https://lnkd.in/g59kP5cp #AI #CleanEnergy #DataCenters #Sustainability #Innovation #GoogleCloud #Infrastructure

AI infrastructure is evolving—and so is how we power it. With Project Jupiter, Oracle Cloud Infrastructure is working with Bloom Energy to bring large-scale fuel cell technology into a microgrid-powered data center campus built for next-generation AI workloads. What stands out isn’t just the fuel cell technology, but the architecture. Integrating onsite generation at this scale allows us to design for performance and resilience from the start, while also reducing emissions and significantly minimizing water use. It’s a different way of thinking about cloud infrastructure, one that reflects both the demands of AI and the importance of building datacenters responsibly. Read more here. https://lnkd.in/dYE5bdvj

Deloitte calls India becoming Asia-Pacific's data centre hub a "rare structural opportunity." Having operated across energy and AI, I agree. India consumes 20% of the world's data and hosts less than 5% of the world's data centres. Lower costs, competitive tariffs, a deep AI talent pool, and a Budget that just proposed a tax holiday until 2047 for foreign cloud companies operating from India. The fundamentals are genuinely strong. But what caught my attention in the report was that the constraint is not demand or capital. It is power. Too often we see everyone talking about AI models. Very few talk about the electricity to run them. Deloitte estimates India's AI data centres will need 40–45 TWH of extra power by 2030. That is a staggering number. And the infrastructure to deliver it does not exist yet. What most people miss is that AI workloads are nothing like traditional IT. GPU clusters need high-density, uninterrupted power with advanced cooling. The existing grid was simply not designed for this. That is where renewables enter the conversation, and where the conversation has to get more honest. We have made serious progress on renewable capacity in India, but generation alone does not solve it. You need transmission upgrades, grid stability, and reliable baseload supply. Without that, you end up with data centres that cannot run at full capacity. I have seen that every hyperscaler conversation eventually comes back to 1 question: can you guarantee clean, reliable power at scale? That is the real qualification test for India right now. The AI value chain does not start with an algorithm. It starts at the power plant. Story by Fortune India: https://lnkd.in/gU4_RsK6