We Build the Data Infrastructure Africa’s Grid Cannot Function Without
8,294 MW of behind-the-meter solar in South Africa alone. 19.3 GW of registered private generation. Zero unified data layer. That’s the gap Asoba closes.
The Problem
Every Major OEM Ships a Different Data Format. On Purpose.
Distributed renewable energy has scaled faster than the software built to manage it. South Africa’s behind-the-meter solar fleet now exceeds the entire grid-tied utility-scale renewable fleet. NERSA has registered 19.3 GW of private generation. None of it speaks the same data language.
Hardware vendors gate telemetry behind proprietary APIs, rate-limited cloud interfaces, and formats that change without notice. Engineering teams spend weeks in parser hell instead of running portfolios. Settlement is done in spreadsheets. Yield loss goes unflagged. The Durban 2.31 MW site in our published research was losing R1.1M per year — undetected, until we ran the audit.
- Inverter API fragmentation: Huawei, SolarEdge, Fronius, Enphase — each OEM ships a different schema, a different refresh rate, a different authentication model. There is no standard.
- Manual settlement: As SAWEM opens private wheeling markets, reconciling Eskom AMR half-hour profiles against generation is still done by hand. That is not a process. That is exposure.
- Reactive operations: Without continuous, normalized telemetry, operators learn about faults from client complaints, not anomaly detection. The financial cost is compounding.
Global solar underperformance losses exceed $10 billion annually. The root cause is not hardware. It’s the absence of an intelligence layer.
The Stack
Open Protocol. Commercial Platform. Research Engine.
Open Protocol
ODSE
Apache-licensed open standard for energy asset data interchange. One schema. Every OEM. pip install odse and your telemetry is normalized in minutes, not weeks.
Commercial Platform
Ona
Portfolio monitoring, JEPA-based fault detection (57.8x false alarm reduction), day-ahead forecasting, and automated wheeling settlement — built on ODSE, delivered as multi-tenant SaaS.
Research Engine
Nehanda & Zorora
Nehanda is a 27B model fine-tuned for epistemic RAG synthesis — 82.21% on FACTS Grounding, outperforming GPT-4o. Zorora runs it against 80+ live market series to produce cited, defensible intelligence reports.
Our Strategy
Platform Formation and Research
The energy transition creates a market for intelligence. Every operator, trader, and regulator in a restructuring grid needs to make decisions from data — and the quality of those decisions is increasingly the difference between a profitable asset and an undetected loss. That is true in Southern Africa. It is true in every market where distributed generation is outpacing the software built to manage it.
Asoba’s position in that market is built on two deliberate choices. The first is to own the data layer before it ossifies around proprietary standards. Open protocols win in fragmented hardware markets — the alternative is a decade of parser hell while the market waits for a vendor to impose lock-in. The second is to treat research as an operating discipline, not a marketing exercise. The architectural questions that determine whether AI output is trustworthy in industrial settings are genuinely hard, and the answers are not obvious. Asoba publishes the work because the rigor is the product.
Those two choices compound. The open standard expands the network. The research earns the trust. Neither works without the other.
Operations
Cape Town. SAPP Region. Global.
Headquartered in Cape Town with active deployments across South Africa and Zimbabwe, and research partnerships in the United States. We operate where the grid is moving fastest and the data infrastructure is most broken — which, right now, means Southern Africa.
Backed By
