CLAI Ventures Invests in Carbon Signal: Building Energy Intelligence at Scale
While AI is typically associated with an abundance of data, it has tremendous potential in industries where the data is sparse, but the underlying physics is well understood. At CLAI Ventures, we believe that the role of AI in these situations is to produce precision that neither the limited data nor the meticulous physics could achieve on its own. Nowhere is this truer than in the global effort to decarbonize buildings, a sector that accounts for roughly 40% of global energy consumption and one-third of all CO₂ emissions.
Today, we are announcing our investment in Carbon Signal, a New York-based company that has built what we believe is the most technically rigorous, enterprise-grade platform for building energy intelligence in the world. While the building data is sparse, often just monthly utility bills and basic metadata, the physics-based energy models are well defined. Carbon Signal’s breakthrough is using AI to bridge that gap: combining limited data with deep physical knowledge to produce decision-grade intelligence that was previously available only through expensive, manual expert analysis.
The Problem: A Ten Trillion-Dollar Decision Gap
The path to net-zero runs directly through buildings. Buildings sit at the center of more than $10 trillion in annual financial, operational, and policy decisions: acquisitions, loans, leases, insurance, retrofits, utility programs, and public regulation. In every one of these decisions, energy performance increasingly affects value, risk, and cost.
The most immediate and capital-intensive of these is the retrofit opportunity: McKinsey, the IEA, and the UNEP GlobalABC collectively estimate that the global building stock will require approximately $2 trillion per year in retrofit capital to meet climate commitments. That wave of spending is arriving now and it requires decision confidence that the industry cannot currently deliver at scale.
Making good retrofit decisions is really hard. Every building is a snowflake. Its energy performance is determined by a unique combination of mechanical systems, envelope characteristics, occupancy patterns, and operational history. Until recently, anyone who wanted a rigorous, trustworthy energy analysis of a building had two choices and neither was adequate at scale:
1. Expert consultant audits: Trusted, physics-grounded, and defensible but they cost tens of thousands of dollars per building and take 6-8 weeks. At portfolio scale, they are simply unaffordable.
2. Data-driven black boxes: Fast and cheap, but opaque, fragile, and difficult for enterprise clients or their boards to defend when allocating serious capital.
This tradeoff — confidence or scale — has blocked capital allocation, slowed retrofit approvals, and misallocated billions in investment. Carbon Signal was built to eliminate it.
The Solution: Physics-Based Intelligence, Born at MIT
Carbon Signal’s core technology was born at MIT’s Sustainable Design Lab, one of the world’s premier research centers for building performance simulation and urban energy modeling, led by Professor Christoph Reinhart.
Shreshth Nagpal, Carbon Signal’s founder and CEO, completed his doctoral research under Professor Reinhart, developing the probabilistic inference methods that became the foundation of the platform. The team carries over a decade of academic research and industry experience across enterprise portfolios.
The core innovation is a physics-constrained inference engine. Here is how it works:
Starting from sparse inputs — primarily monthly utility bills and basic building metadata — the engine applies thermodynamic principles, system logic, and occupancy schedules to constrain the solution space.
Rather than producing a single deterministic answer (which would be false precision), it generates an ensemble of feasible building energy models, each consistent with the available data and physical constraints.
Outputs are presented as probability ranges, not point estimates, enabling capital planning decisions with explicit awareness of uncertainty.
The result: audit-grade analysis in 5–10 minutes, at a fraction of the cost of traditional approaches. The process that used to take 6-8 weeks and tens of thousands of dollars per building can now be run across thousands of buildings in a single afternoon.
Traction: Enterprise Validation Across Sectors and Continents
Carbon Signal’s platform has been adopted by some of the world’s most demanding enterprise organizations: Mag 7 big tech, leading academic institutions like MIT, Yale, Georgia Tech, and Johns Hopkins, prominent Real Estate Investment Trusts (REITs) and municipal and governments. It’s used in sectors spanning logistics, commercial offices, grocery, healthcare, higher education, retail, and industrial real estate across North America, Europe, the Middle East, and Australia. The platform has been deployed across 10,000+ buildings, with over 3,000 buildings under active subscription.
A few highlights from our diligence:
Adopted, renewed, and expanded by the world’s most demanding enterprise clients across multiple sectors and geographies.
$600M+ in retrofit capital decisions have been informed by Carbon Signal analysis.
3M+ metric tons of CO₂ reduction roadmaps have been developed on the platform.
In one validated case study, Carbon Signal’s predicted energy savings for an MIT building matched actual implemented savings closely after a retro-commissioning project which gives a hard technical validation of the engine’s accuracy.
The platform has been independently adopted by multiple divisions of the same global enterprise, each division discovering it separately, on merit, a powerful signal of genuine product-market fit.
The platform delivers analysis at 10–20x lower cost than traditional energy audits, compressing monthslong processes into minutes.
Critically, demand is durable regardless of shifts in federal climate policy because of state and local regulations in the US and EU building performance regulations in Europe. State and local regulations (NYC Local Law 97, California AB 802 benchmarking fees, Maryland’s Building Performance Standards covering 10,000+ buildings) create enforceable demand floors independent of Washington.
The Team: Where Building Science Meets Startup Execution
The founding team combines rare depth in building science and AI with real-world enterprise implementation experience.
Shreshth Nagpal (CEO) holds a PhD from MIT’s Sustainable Design Lab and brings 20+ years of building science experience.
The broader team includes three MIT PhDs and post-doctoral researchers, covering machine learning, building technology, and thermodynamic modeling, alongside engineers and customer success leads who bring prior startup exits and deep real estate sustainability consulting experience.
The unique aspect of the product is that it was built in close collaboration with one of the largest big tech clients, and refined through near weekly conversations with their sustainability teams managing portfolios of thousands of buildings.
The Vision: From Application to Infrastructure
Carbon Signal’s long-term vision is to move from application to infrastructure, a trajectory that mirrors how Google Maps evolved from a mapping application into infrastructure that powers ride-sharing, logistics, and delivery.
Today, the platform serves as a portfolio intelligence application for real estate owners, investment managers, and their advisors. The future state is powering decision ecosystems by embedding into capital planning systems, compliance platforms, lender underwriting tools, and advisory workflows.
The more deeply embedded the platform becomes in enterprise workflows, the higher the switching cost. As compliance platforms, lending programs, and advisory practices build on Carbon Signal’s intelligence layer, the platform becomes structural rather than discretionary.
Why We Invested
At CLAI Ventures, we invest at the intersection of AI and climate, backing companies where software intelligence creates measurable, scalable, and durable impact. Carbon Signal hits every dimension of that thesis:
Technical moat: A physics-constrained inference engine born at MIT, backed by 10+ years of academic research. No identified competitor combines building-specific physics inference with quantified uncertainty at software scale.
Enterprise validation: Adopted, renewed, and expanded by the world’s most demanding enterprise clients across multiple sectors and geographies. Zero churn.
Revenue quality: High gross margins and a pipeline of near-term revenue catalysts across new verticals and geographies.
Durable demand: Regulatory compliance at the state and local level, and EU regulations create real financial incentives.
Multi-stakeholder leverage: The same intelligence layer serves owners, consultants, lenders, utilities, and compliance platforms, creating multiple revenue streams per building rather than a single subscription.
We are proud to support Shreshth and the Carbon Signal team as they scale a building energy intelligence layer that delivers decision confidence across the global built environment transition.
If you are an investor interested in learning more we would be happy to make an introduction. Reach us at ajay@clai.vc and team@clai.vc.
This post was written by Ajay Gupta and the CLAI Ventures team, a Silicon Valley-based fund investing at the intersection of AI and climate.