Zeitview

Learning Objectives

• Understand how AI-assisted aerial inspection works for solar farms and wind turbines, and where the computer vision actually sits in the workflow.
• Learn why a published precision figure of roughly 68 percent makes human review an architectural requirement rather than a temporary crutch.
• Evaluate Zeitview honestly — separating the rigorously evidenced wind blade research from the marketing-only claims around its solar products.

What Is Zeitview?

Zeitview is AI-assisted aerial inspection software and services for solar, wind, utility, telecom, and property assets. Drone and crewed-aircraft imagery is run through computer-vision models, and human analysts finalize the reports. The problem it addresses is physical and unglamorous: a utility-scale solar farm has hundreds of thousands of panels and a wind farm has blades over one hundred feet long, and sending technicians to inspect each one by hand is slow, expensive, and dangerous. Flying the asset and letting a model triage the imagery is a genuinely better starting point.

The company was founded in 2014 and is headquartered in Santa Monica, California. It launched as DroneBase and rebranded to Zeitview on February 7, 2023, alongside a 55 million dollar raise led by Valor Equity Partners — the name dropped "drone" because the business had expanded beyond drones into crewed aircraft and ground capture. Founder and chief executive Dan Burton is a former Goldman Sachs employee and a former Marine infantry officer, and the company is a Y Combinator alum. A further 60 million dollars arrived in March 2025, led by Climate Investment, with Union Square Ventures, Upfront Ventures, Energy Transition Ventures, and Y Combinator among the backers. The company remains private with no public offering. Total funding raised is disputed across trackers, and no valuation has been verified, so neither figure is stated here.

Core Capabilities

Solar Insights — thermal and infrared anomaly detection

Solar Insights handles photovoltaic inspection, using thermal and infrared imagery to detect anomalies including hotspots, diode failures, string outages, and electrical isolations. These are the failure modes that quietly erode a solar farm's output without triggering any alarm at the inverter. Important caveat: no architecture, accuracy figures, or dataset detail have been published for the solar and thermal models. The capability is asserted in marketing only.

Wind Insights — blade damage over time, including internal inspection

Wind Insights tracks turbine blade damage across repeat inspections, so operators can see whether a crack is stable or propagating. The genuine differentiator here is internal blade inspection, performed using caged drones and crawlers that fly or drive inside the blade cavity. External blade capture is a competitive field; the interior is much less contested.

The published crack-detection research and its pipeline

Zeitview maintains an in-house AI Research team, which published a peer-reviewed paper in 2024 on detecting barely-visible surface cracks on wind-turbine blades. The team built a proprietary dataset of 9,107 high-severity cracks drawn from 4,684 turbine inspections across 988 locations, 34 manufacturers, and 195 turbine models — real inspection imagery, not simulated. They explicitly criticized prior datasets in the field for being simulated or augmented. Models tested included ResNet-18, EfficientNet-B3, and MobileNetV3.

The deployed pipeline runs in four stages, and each stage is worth understanding because it is a template for how this kind of system is genuinely built.

The drone-pilot capture network

Zeitview is a managed service, not a self-serve product. Customers submit inspection requests, Zeitview converts them into standardized mission templates, and local gig-economy drone pilots claim and fly the missions through the Zeitview Pilot app. This is how the company achieves geographic coverage without maintaining its own flight crews everywhere, and it is also why the business is best described as hybrid services and software rather than a pure AI product.

North American Solar Scan

The North American Solar Scan rates large United States solar assets on a standardized scale, producing a comparable view across the fleet rather than a single site report. Zeitview states it captured 95 gigawatts of the United States solar fleet in a 2023 scan.

Strengths

• Published, peer-reviewed evidence for its wind blade AI — a proprietary dataset of real inspection imagery, named model architectures, and honest metrics. This is rare in industrial computer vision.
• Internal blade inspection via caged drones and crawlers is a genuine capability that most competitors do not offer.
• Broadest horizontal coverage — solar, wind, utility, telecom, and property assets in one platform, rather than a single-vertical specialist.
• Named customers on its own site including Vestas, Siemens Gamesa, EDF Renewables, Duke Energy, CBRE, and Hines, plus Bureau Veritas North America and Nexamp case studies.
• Well capitalized for the sector, with 60 million dollars raised in March 2025 led by Climate Investment.
• An honest research posture — the team criticized simulated datasets in prior work and published metrics that do not flatter their own product.

Limitations and Considerations

• Precision of roughly 68 percent means about one in three flagged cracks is a false positive. Budget for analyst review time; this is not an autonomous inspector.
• The rigorous published evidence covers wind blade cracks only. Do not extrapolate the paper's credibility to Solar Insights — the solar and thermal AI has no published architecture, accuracy figures, or dataset detail.
• It is a hybrid services and software business. Gig drone pilots capture the imagery and in-house analysts verify the defects. Data labeling was outsourced: a contractor labeled 37,000 aerial inspection images over 15 months. That labeling is normal, legitimate supervised-learning practice and actually confirms real model training — but it shows the human labor footprint is large.
• Imagery quality dependence is acknowledged by Zeitview itself. The models need diverse, well-annotated data across defect types and environmental conditions, and infrared normalization is a problem the team discusses without claiming to have solved.
• Barely-visible hairline cracks are genuinely hard. That difficulty is the paper's whole premise, and the modest metrics reflect it rather than indicating a weak team.
• There is a marketing-versus-substance gap. The website says "AI-powered" everywhere with no metrics; the research paper has metrics and is far more modest. Trust the paper.
• Scale claims are vendor-reported. About 200,000 assets inspected across 80 countries in 2024, and 95 gigawatts of the United States solar fleet captured in the 2023 scan, are Zeitview's own figures.
• Employee reviews cite two rounds of layoffs — a small sample, so treat it as directional only. Recent output also shows a visible pivot toward property and insurance, which dilutes the renewables-first framing.

Best Use Cases

Getting Started

1. Scope the asset class honestly. If your priority is wind blade condition, Zeitview's evidenced strength lines up with your need. If it is solar-only depth, evaluate Raptor Maps alongside it; if it is external wind capture autonomy, evaluate SkySpecs.
2. Submit an inspection request. Zeitview converts it into a standardized mission template that local pilots in its network claim and fly through the Zeitview Pilot app.
3. Plan for analyst review in your workflow, not around it. With precision of roughly 68 percent on the published crack model, the deliverable is a triaged shortlist that humans have already filtered — build your maintenance planning on the reviewed report.
4. Ask for the metrics behind whichever product you are buying. The wind blade research is public and specific. For the solar and thermal side, ask the vendor directly for dataset size, defect-type coverage, and accuracy on your environmental conditions.

Key Takeaways

• The genuine AI in renewables is perception — computer vision that sees a cracked blade or a dead panel from the air — while physics engines do the accuracy. Zeitview sits squarely on the perception side.
• Zeitview's published wind blade crack detector reached about 82 percent accuracy with precision of roughly 68 percent, on a proprietary dataset of 9,107 real cracks from 4,684 inspections. Publishing those numbers at all sets it apart in a category where most vendors publish none.
• Precision of roughly 68 percent means human-in-the-loop review is architecturally required, not a marketing footnote. The model proposes regions; analysts decide.
• The rigorous evidence covers wind blades only. The solar and thermal AI is asserted in marketing with no published metrics — evaluate the two products separately.
• Competitively, Zeitview is the broad horizontal player against Raptor Maps in solar depth and SkySpecs in wind, and its sharpest counter-differentiator is internal blade inspection via caged drones and crawlers.