Case study: Climate risk stress testing & scenario regulation — a startup-to-enterprise scale story

When Jupiter Intelligence closed its $54 million Series C round in late 2023, the climate analytics startup had grown from a three-person academic spinout to an enterprise platform serving over 100 institutional clients across banking, insurance, and real estate. The company's trajectory from early prototype to enterprise-grade climate risk stress testing platform illustrates both the enormous demand for physical climate risk analytics and the operational hurdles that separate promising technology from bankable infrastructure. A 2025 survey by the Network for Greening the Financial System (NGFS) found that 78% of central banks and financial supervisors now require or plan to require climate scenario analysis from regulated institutions, up from 31% in 2021. For founders building in this space, the path from climate model to enterprise contract is neither linear nor simple.

Why It Matters

Climate risk stress testing has shifted from a voluntary exercise to a regulatory mandate across major financial markets. The European Central Bank's 2022 climate stress test covered 104 significant banks representing EUR 24 trillion in assets and revealed that most institutions lacked the granular, forward-looking physical risk data needed to meet supervisory expectations (ECB, 2022). In the United States, the Federal Reserve's pilot climate scenario analysis in 2023 engaged the six largest US banks and highlighted critical gaps in the translation of climate science into financial risk metrics. The Bank of England's biennial exploratory scenario exercises, running since 2021, have progressively tightened expectations around scenario granularity, asset-level exposure mapping, and transmission channel modeling.

The regulatory pressure creates a structural market opportunity. McKinsey estimated the global climate risk analytics market at $3.2 billion in 2024, growing at 28% annually through 2030. However, the technical requirements are formidable: enterprise clients need asset-level physical risk projections at sub-kilometer spatial resolution, across multiple climate scenarios (typically NGFS orderly, disorderly, and hot-house pathways), with probabilistic output suitable for integration into existing risk management frameworks. Building this capability requires combining climate science, geospatial analytics, financial modeling, and enterprise software engineering, a combination that very few teams possess.

Key Concepts

Climate risk stress testing evaluates how financial institutions, real estate portfolios, or infrastructure systems would perform under specified climate scenarios. Physical risk analysis focuses on hazards such as flooding, wildfire, extreme heat, hurricanes, and sea-level rise. Transition risk analysis examines exposure to policy changes, carbon pricing, technology shifts, and market sentiment changes.

Scenario analysis uses standardized climate pathways, most commonly the NGFS scenarios or the Representative Concentration Pathways (RCPs) from the Intergovernmental Panel on Climate Change, to project future hazard conditions. Stress testing applies these scenarios to specific portfolios or balance sheets to estimate financial impacts including expected losses, capital adequacy implications, and asset valuation changes.

The technical stack typically includes downscaled general circulation model (GCM) output, hazard-specific catastrophe models, exposure databases mapping assets to geographic coordinates, and vulnerability functions that translate physical hazard intensity into financial loss estimates. Enterprise integration requires APIs, data pipelines compatible with existing risk systems, and audit trails that satisfy regulatory scrutiny.

What's Working

Jupiter Intelligence: From Academic Spinout to Enterprise Platform

Jupiter Intelligence was founded in 2017 by Rich Sorkin (CEO) and a team of climate scientists from academic institutions including MIT and the National Center for Atmospheric Research. The company's initial product, ClimateScore, provided physical climate risk ratings at the asset level using proprietary downscaling of global climate models to 90-meter spatial resolution.

The company's early go-to-market strategy focused on insurance and reinsurance companies, which had existing catastrophe modeling infrastructure and understood probabilistic risk assessment. Jupiter secured its first enterprise contracts with Munich Re and Swiss Re in 2019, providing supplementary peril analytics for flood and wildfire hazards that traditional catastrophe models underestimated under climate change scenarios. These initial contracts, valued at $500,000 to $1.5 million annually, validated product-market fit but also revealed critical gaps: insurers needed not just hazard scores but full loss exceedance curves integrated into their existing underwriting workflows.

Jupiter's pivot to a platform model in 2020 to 2021, offering FloodScore, FireScore, HeatScore, and WindScore as modular API products, enabled integration with enterprise risk systems including Moody's RiskFirst, S&P Global Market Intelligence, and Bloomberg Terminal. This architectural decision proved pivotal: by 2023, Jupiter reported that 85% of revenue came through platform integrations rather than standalone reports, with average contract values increasing to $2 million to $5 million annually.

The company raised $134 million in total funding through 2024, with investors including Energize Ventures, American Family Insurance Institute for Corporate and Social Impact, and QBE Ventures. Jupiter's client base expanded from insurance to banking (serving five of the six banks participating in the Fed's pilot climate scenario analysis), real estate investment trusts, and government agencies including FEMA and the US Army Corps of Engineers.

Moody's Analytics: Acquisition as Scale Strategy

Moody's acquisition of RMS (Risk Management Solutions) for $2 billion in 2021 illustrates the acquisition pathway to enterprise scale in climate risk analytics. RMS had spent over a decade building catastrophe models for the insurance industry and had begun extending these models to incorporate climate change projections. Post-acquisition, Moody's integrated RMS climate analytics into its broader credit risk and ESG assessment platforms, enabling a bundled offering that covers both physical and transition risk.

The integration allowed Moody's to offer climate-conditioned probability of default (PD) and loss given default (LGD) estimates directly within its credit risk models, a capability that banks needed for regulatory stress testing but could not easily build internally. By 2025, Moody's reported that over 200 financial institutions used its climate risk modules, with particular adoption in Europe where ECB supervisory expectations were most prescriptive.

Cervest: Geospatial Intelligence for Real Assets

Cervest, a London-based climate intelligence startup, took a different approach by focusing on real asset owners and operators rather than financial institutions. Founded in 2016, Cervest built its EarthScan platform to provide asset-level climate risk ratings for commercial real estate, infrastructure, and agricultural land. The platform combined satellite imagery, IoT sensor data, and climate model output to generate dynamic risk scores updated in near-real-time.

Cervest raised $37 million through 2023 and secured clients including935 Group,935 Properties, and several UK pension funds responding to Taskforce on Climate-related Financial Disclosures (TCFD) requirements. The company's emphasis on visualization and non-technical user interfaces enabled adoption by sustainability teams and board members who lacked the technical background to interpret raw climate model output.

What's Not Working

Model Validation and Regulatory Acceptance

The most persistent challenge for climate risk startups is model validation. Unlike traditional financial models that can be backtested against historical data, climate risk models project conditions 10 to 50 years into the future under scenarios that have no historical precedent. Regulators, particularly the ECB's Supervisory Review and Evaluation Process (SREP) teams, have pushed back on climate risk assessments that lack transparent validation methodologies. A 2024 review by the Basel Committee on Banking Supervision found that fewer than 20% of banks could demonstrate rigorous validation of their climate scenario models, citing "black box" vendor models as a primary concern (BCBS, 2024).

For startups, this creates a paradox: enterprise clients need validated models to satisfy regulators, but validation frameworks for forward-looking climate risk are still under development. Jupiter addressed this by publishing peer-reviewed validation studies comparing its hazard projections against observed extreme weather events and by participating in the NGFS climate scenario validation working group. However, this investment in scientific credibility required sustained R&D spending that compressed margins during the critical growth phase.

Data Quality and Granularity Gaps

Asset-level climate risk assessment requires precise geolocation of physical assets, detailed structural characteristics (construction type, elevation, flood defenses), and exposure values. Many financial institutions lack this data for their loan portfolios. A 2025 survey by the European Banking Authority found that 62% of banks could geolocate fewer than half of their commercial real estate exposures to specific building addresses, relying instead on postal code or city-level approximations that introduce substantial error in flood and wildfire risk estimates (EBA, 2025).

Startups that assumed clients would provide clean, geocoded asset data discovered that significant professional services effort was required to cleanse, geocode, and enrich client portfolios before analytics could be applied. This services burden slowed deal cycles, reduced gross margins, and created scaling challenges that pure software economics did not anticipate.

Scenario Standardization Fragmentation

Financial institutions operating across multiple jurisdictions face a patchwork of scenario requirements. The ECB mandates NGFS scenarios; the Bank of England uses its own bespoke Climate Biennial Exploratory Scenario (CBES) pathways; the Fed's pilot used a distinct set of physical and transition scenarios; and the Hong Kong Monetary Authority requires scenarios calibrated to Asian climate conditions. Startups building products for global banks must support multiple scenario frameworks simultaneously, multiplying development and maintenance costs.

Key Players

Established: Moody's Analytics (integrated climate-credit risk platform post-RMS acquisition), S&P Global (Climate Credit Analytics with physical and transition risk modules), MSCI (Climate Value-at-Risk covering over 10,000 companies), Bloomberg (climate scenario analysis tools integrated into Terminal), Willis Towers Watson (Climate Quantified advisory and analytics practice)

Startups: Jupiter Intelligence (asset-level physical risk analytics with modular API platform), Cervest (EarthScan geospatial climate intelligence for real assets), Intensel (AI-powered physical climate risk for real estate and infrastructure), ClimateAI (supply chain climate risk analytics), Sust Global (satellite-derived physical risk data and APIs)

Investors: Energize Ventures (Jupiter Intelligence lead investor), QBE Ventures (strategic insurance industry investor), Lightspeed Venture Partners (climate analytics portfolio), Generation Investment Management (climate finance ecosystem), American Family Insurance Institute for Corporate and Social Impact (Jupiter Intelligence investor)

Action Checklist

• Identify your beachhead market segment: insurance, banking, real estate, or government each have distinct data requirements, sales cycles, and regulatory drivers
• Build API-first architecture from day one to enable integration with enterprise risk platforms rather than standalone report delivery
• Invest in peer-reviewed model validation studies before enterprise sales conversations, as regulatory credibility accelerates deal closure
• Plan for significant data cleansing and geocoding services in early client engagements and price these into contracts rather than absorbing them
• Support multiple scenario frameworks (NGFS, CBES, Fed, HKMA) to serve global financial institutions without requiring custom development per client
• Establish partnerships with incumbent risk data providers (Moody's, S&P, Bloomberg) as distribution channels rather than competing for direct enterprise relationships
• Maintain a climate science advisory board with peer-reviewed publication track records to build credibility with both regulators and sophisticated institutional buyers
• Budget for 18 to 24 month enterprise sales cycles in banking and insurance, with proof-of-concept phases consuming 4 to 6 months before contract execution

FAQ

Q: How long does it typically take a climate risk analytics startup to close its first enterprise banking contract? A: Based on the trajectories of Jupiter Intelligence, Cervest, and comparable companies, the first enterprise banking contract typically takes 12 to 24 months from initial engagement to signed agreement. The process involves a technical due diligence phase (2 to 4 months), a proof-of-concept on a sample portfolio (3 to 6 months), procurement and legal review (2 to 4 months), and model validation review by the bank's internal risk team (2 to 4 months). Startups that enter with a reference client from the insurance sector can compress this timeline by 3 to 6 months because insurers and banks share regulatory frameworks and risk assessment vocabulary.

Q: What technical capabilities are non-negotiable for serving regulated financial institutions? A: Regulated institutions require: asset-level physical risk projections at sub-kilometer resolution across at least three NGFS-aligned scenarios and two time horizons (2030 and 2050 minimum); probabilistic output (not just deterministic scores) compatible with Value-at-Risk and expected loss frameworks; transparent model documentation sufficient for regulatory model risk management review under SR 11-7 or equivalent; API-based delivery with SOC 2 Type II security certification; and audit trails tracking data lineage from raw climate model input through to final risk metric output.

Q: Is it better to build a full-stack climate risk platform or specialize in a single peril or asset class? A: The evidence from successful scale-ups suggests starting with one or two perils in a single asset class, then expanding. Jupiter Intelligence began with flood and wildfire risk for insured properties before adding heat stress, wind, and drought and expanding into banking and real estate. This focused approach allowed the company to achieve depth of validation and operational reliability in its initial capabilities before broadening. Companies that attempted full-stack coverage from launch, including several that did not survive past Series A, found that the quality demands of enterprise buyers and regulators exceeded their capacity to deliver across multiple perils simultaneously.

Q: How do climate risk analytics companies handle the tension between scientific uncertainty and client demands for precise numbers? A: Successful companies communicate uncertainty explicitly through probability distributions, confidence intervals, and scenario ranges rather than single-point estimates. Jupiter Intelligence delivers results as probability distributions (e.g., "5th to 95th percentile flood depth range under RCP 8.5 in 2050") and provides documentation explaining the sources and magnitudes of uncertainty. This approach initially met resistance from clients accustomed to deterministic credit ratings but has become the expected standard as regulatory guidance has increasingly emphasized the importance of representing uncertainty in climate risk assessments.

Q: What is the role of open-source climate data versus proprietary models in building a defensible business? A: Most successful climate risk analytics companies use freely available climate model output (CMIP6 ensemble) and public hazard datasets as foundation layers, then add proprietary value through statistical downscaling algorithms, calibration against observed data, hazard-specific engineering models, and financial loss translation functions. The open-source climate data is necessary but not sufficient: the defensible intellectual property lies in the translation from raw climate projections to financially actionable risk metrics at asset-level resolution.

Sources

• European Central Bank. (2022). 2022 Climate Risk Stress Test: Main Results and Key Takeaways. Frankfurt: ECB Banking Supervision.
• Network for Greening the Financial System. (2025). Climate Scenario Analysis by Jurisdictions: Progress and Challenges. Paris: NGFS Secretariat.
• Basel Committee on Banking Supervision. (2024). Climate-Related Financial Risk: Progress in Supervisory Approaches and Bank Practices. Basel: Bank for International Settlements.
• European Banking Authority. (2025). Report on the Role of Environmental and Social Risks in the Prudential Framework. Paris: EBA.
• McKinsey & Company. (2024). Climate Risk Analytics: Market Sizing and Growth Projections 2024-2030. New York: McKinsey Global Institute.
• Federal Reserve Board. (2023). Pilot Climate Scenario Analysis Exercise: Summary of Preliminary Results. Washington, DC: Board of Governors of the Federal Reserve System.
• Jupiter Intelligence. (2024). ClimateScore Platform: Validation Report and Methodology Documentation. San Mateo, CA: Jupiter Intelligence Inc.
• Bank of England. (2024). Results of the 2023 Climate Biennial Exploratory Scenario. London: Bank of England Prudential Regulation Authority.