Candor doesn't just measure emissions — it helps you reduce them by aligning workloads with real-time grid conditions. Total footprint, AI scope, workload scheduling recommendations — grid-matched, not RECs.
The world's largest technology companies are spending billions on Renewable Energy Certificates while their actual cloud emissions climb year over year — and AI workloads are accelerating the curve. The problem isn't a lack of renewable energy investment; it's a lack of real-time, independent measurement tying that investment to actual workload consumption.
Real-time grid carbon intensity matched to your actual cloud workloads.
AI scope, general compute, and LLM usage broken out separately — no blending, no hiding.
Schedule workloads for cleaner windows. Candor shows you when and where to run for lower carbon.
A company can purchase RECs from a wind farm in Wyoming to cover a data center in Virginia running on coal. No temporal matching. No geographic matching. No connection to reality.
— The REC credibility gap, explained
The industry is catching up. The GHG Protocol is evolving its Scope 2 guidance. The emerging 24/7 Carbon Free Energy standard — which calls for hourly grid matching — is becoming the benchmark serious enterprises hold themselves to. Location-based methodology, applied in real time, is the only honest answer.
Cloud emissions are reported as Scope 3, but they increasingly behave like Scope 2.
As AI workloads scale, compute is becoming a primary driver of emissions — yet companies still rely on coarse estimates and annual reconciliation methods that are not tied to when or where workloads actually run.
At the same time, the Greenhouse Gas Protocol is evolving Scope 2 guidance toward time- and location-based accounting, including hourly matching and stricter geographic alignment. These expectations are beginning to extend into how companies are expected to understand cloud emissions.
Candor brings Scope 2-level visibility to cloud workloads — making it possible to understand when compute was actually clean, and to act on it.
This initiative establishes a framework for enterprises operating cloud workloads — with explicit AI-scope segmentation — to measure, not estimate, the electricity behind their compute at the moment of execution. The approach is grounded in four principles:
Rather than relying on annual REC reconciliation, this framework connects enterprise cloud infrastructure directly to live grid carbon intensity signals — producing a timestamped record of how clean the grid was when each workload ran, with AI-scope broken out for Scope 2-aligned electricity tracking.
The initiative targets enterprises that have made public net-zero commitments and operate significant cloud infrastructure — particularly those with meaningful AI workload exposure and a stated interest in 24/7 Carbon Free Energy accounting.
The core claim is deliberately conservative: grid cleanliness at the time of workload execution — aligned with GHG Protocol location-based methodology and the emerging 24/7 CFE standard. We do not redefine Scope 2; we make the electricity behind cloud workloads visible.
Candor turns cloud electricity from a reporting problem into an operational decision.
Every cloud workload is matched against live Marginal Operating Emissions Rate (MOER) data from WattTime — a leading signal for location-based electricity accounting. No annual averages. No REC reconciliation.
GPU instances are segmented from general compute inside the total cloud footprint. LLM API usage is labeled separately as modeled estimates — never mixed with calculated cloud infrastructure data. Enterprises get an honest AI-scope breakdown, not a blended average.
Measurement is derived independently from cloud provider disclosures. Every workload is logged to an append-only, timestamped record with region, instance type, energy estimate, and verified grid signal. Every data point is traceable end-to-end.
14-day historical grid patterns surface low-carbon windows per region and drive scheduling recommendations for AI training jobs, batch inference, and deferred compute. Measurement becomes a continuous improvement loop — visibility that drives action.
Enterprise connects AWS, Azure, or GCP accounts via read-only credentials. No write access, no operational risk. CloudWatch and billing APIs pull instance-level data.
GPU instances are identified by type (p4, p3, g5 on AWS; NC/ND on Azure; TPUs on GCP) and segmented from general compute. The AI scope is tracked and reported in parallel with the full cloud footprint, never replacing it.
Each workload's execution window is matched against WattTime's verified real-time MOER signal for the corresponding grid region. TDP-based energy figures — industry-standard methodology — are clearly labeled as estimates, while the grid signal driving them is live and independently verified.
Every workload is written to an append-only, timestamped database record: org, timestamp, cloud provider, region, instance type, hours, kWh, carbon intensity, CO₂ (kg), and renewable percentage.
Export-ready, Scope 2-aligned reporting for cloud and AI workloads, supporting sustainability disclosures, internal reporting, and 24/7 carbon-free energy (CFE) tracking. AI and total cloud emissions are separated; LLM usage is transparently modeled and distinguished from calculated infrastructure.
Candor's output is aligned with — not a replacement for — the frameworks enterprises are using to account for cloud electricity. The methodology is conservative by design: we measure what actually happened on the grid when your workloads ran, and we leave classification and reporting decisions to your sustainability team.
The emerging 24/7 Carbon Free Energy standard calls for hourly grid matching — moving beyond annual REC reconciliation toward genuine real-time accountability. Enterprises that establish real-time electricity measurement practices today build a durable foundation for carbon-aware compute, regardless of how future methodology evolves.
WattTime real-time MOER signal active on CAISO_NORTH. First cloud carbon tracking reports generated. AWS CloudWatch connection in progress. Backend deployed to Railway.
Full AWS integration with job-level workload tracking. Legal entity formed. First 3–5 pilot customers at discounted rate. WattTime Analyst tier negotiated.
Azure and GCP integrations. Real-time dashboard for enterprise customers. Automated scheduling recommendations. Seed raise initiated.
Established as the independent measurement layer for enterprise cloud electricity — including AI-scope attribution — the Bloomberg of cloud carbon data. Enterprise-grade security review. Series A.
Enterprises with public net-zero commitments and meaningful cloud workload exposure — with AI accelerating the curve — face a widening credibility gap between their REC-based claims and what's actually happening on the grid. We're building the measurement infrastructure to close it.
We're currently in conversations with early enterprise partners, sustainability teams, and climate-focused investors who want to lead on carbon-aware compute.