In a world where AI compute demand is outpacing grid capacity, Emerald AI is tackling one of the most overlooked problems in climate tech: the energy waste of always-on data centers. With its software platform, Emerald Conductor, the company is building a new layer of intelligence that turns energy-hungry data centers into flexible, grid-stabilizing assets, routing compute when and where clean power is available. Backed by NVIDIA and Radical Ventures, and already being piloted in energy-constrained regions, Emerald is positioning itself as a critical software layer in the AI infrastructure stack.
At the core of Emerald AI’s platform is the Emerald Conductor, a real-time orchestration engine that dynamically shifts AI and compute workloads across multiple data center locations. Instead of treating data centers as static infrastructure, the Conductor analyzes grid carbon intensity, power pricing, and network latency in real-time to determine the optimal placement and timing of compute tasks.
The result is a platform that helps reduce emissions, avoid peak demand pricing, and increase overall grid resilience, all without requiring any hardware retrofits. For AI workloads and latency-tolerant batch processing, the optimization can translate directly into lower operating costs and quantifiable Scope 2 emissions reductions. Emerald doesn’t own infrastructure. It sits on top of hyperscale and colocation providers, integrating with existing orchestration tools to quietly make them smarter.
Founders & Team
Emerald AI was founded by a team with deep roots in both cloud infrastructure and grid optimization. The CEO, Jason Selch, previously led strategy at a hyperscale data center firm and brings a pragmatic view of how to scale energy-efficient operations without disrupting uptime. The CTO, Dr. Sarah Lin, holds a PhD in power systems engineering and built predictive models for grid balancing at ISO New England.
What sets this team apart is their ability to speak fluently across two highly specialized domains: cloud orchestration and energy markets. They aren’t trying to force a clean tech thesis into tech infrastructure. They’re building a software product that meets data center operators where they already are, while solving a problem that utilities, regulators, and ESG funds are actively prioritizing.
Funding
Emerald raised a $24.5 million seed round in early 2025. The round was led by NVIDIA’s venture arm and Radical Ventures, with participation from Ampol and several energy transition-focused angels. The raise was designed to support early commercial pilots, expand the engineering team, and deepen integrations with major orchestration frameworks like Kubernetes and Slurm.
What stands out about this cap table is that it aligns commercial incentives across compute and energy. NVIDIA gets visibility into energy-aware infrastructure to support their own chips. Ampol brings access to utility partners where grid balancing is urgent. It’s the kind of investor mix that signals strategy, not just capital.
Key Milestones
- Q1 2025: Closed $24.5M seed round led by NVIDIA and Radical Ventures, with participation from Ampol and Transition VC
- Q2 2025: Launched pilot deployments with data centers in Texas (ERCOT) and Northern California (CAISO), focusing on latency-tolerant AI model training workloads
- Q2 2025: Achieved an average 22 percent reduction in power costs across pilot sites by dynamically shifting compute during low-carbon intensity windows
- Q3 2025 (in progress): Integrating with Slurm, Kubernetes, and NVIDIA Base Command to support broader enterprise adoption
- Q3 2025 (in progress): Early access partnership signed with a West Coast-based hyperscale provider and a global colocation firm (likely Digital Realty or Equinix) to test multi-region routing at scale
- Q4 2025 (planned): Targeting 1 GW of orchestrated compute under management across North America and initial test deployments in Germany and the Netherlands, where grid flexibility incentives are most favorable
Market Opportunity
The intersection of AI infrastructure and energy flexibility is one of the most underexplored but increasingly urgent markets. Global data center power demand is expected to triple by 2030, driven largely by AI training and inference workloads. In the U.S. alone, data centers are already responsible for more than 2.5 percent of total electricity consumption, and that number is climbing fast.
Emerald AI is positioning itself within two overlapping wedges of this trend:
- Energy-Responsive Compute Management: A software market that includes orchestration platforms, smart workload routing, and energy-aware scheduling. No clear category leader has emerged, and infrastructure players are actively looking for solutions.
- Grid-Integrated Software Solutions: As utilities begin to pay large consumers for flexibility, there’s growing value in being a “demand-side partner.” This includes grid-responsive EV charging, HVAC load balancing, and now, intelligent compute.
What makes this a unique wedge is that Emerald isn’t just optimizing for cost. It’s targeting operators who care about energy arbitrage, ESG reporting, and grid-friendly behavior—segments with clear regulatory tailwinds and growing budget allocations.
Competitive Landscape
Most players focused on clean compute fall into one of three camps: hardware retrofit solutions, carbon accounting platforms, or long-duration energy storage. Emerald sits apart by avoiding physical dependencies and by creating value in the orchestration layer.
Competitors:
- Crusoe Energy: Captures stranded gas for on-site compute. Strong for edge compute but limited scalability in structured cloud environments.
- Form Energy: Building iron-air long-duration batteries. Competes indirectly by addressing the same grid inflexibility, but with capex-heavy infrastructure.
- FlexiDAO / Watershed: Focused on clean energy accounting and carbon reporting. Emerald differs by acting before emissions occur, not after.
The closest analog may be AutoGrid, which optimizes flexible industrial loads, but Emerald is going narrower and deeper into compute. That makes it less of a broad platform bet and more of a surgical tool for a very high-value pain point.
Why It Matters
The rise of AI is colliding headfirst with the energy transition. Most people aren’t talking about it yet, but powering LLMs and large compute clusters with 24/7 clean energy is quickly becoming both a technical constraint and a reputational risk for major cloud providers.
Emerald AI is stepping into that tension with a solution that doesn’t require new hardware, new contracts, or new power plants. It’s simply making what already exists smarter. In a capital-constrained environment, that matters.
The broader climate tech world is full of ambitious solutions that depend on massive infrastructure buildouts. Emerald’s advantage is that it can scale without a shovel. For a segment as critical as data center growth, that kind of leverage is rare.
Follow Emerald AI as They Bring Intelligence to the Infrastructure Layer
AI is already transforming how we work, communicate, and build. But behind every model run or chatbot query is a massive energy footprint most people never see. Emerald AI is taking on that invisible problem and offering a practical path forward — one that’s scalable, software-driven, and grid-aware.
If you’re tracking the convergence of AI, clean energy, and infrastructure, this is a company worth watching closely.
Subscribe to The Green Brief for monthly startup spotlights, weekly news recaps, and quarterly market outlooks across clean tech, climate SaaS, and frontier infrastructure.
