TerraWatt

Landing Page
Your Land Plans
Energy Savings

TerraWatt

Inspiration

Solar is finally cheap enough to matter for farms, but the planning layer hasn't caught up. The U.S. has about 879 million acres of farmland, yet only 0.14% is used for solar today, even though DOE scenarios suggest we may need around 10.3 million acres of solar by 2050 which is still under 1.2% of existing farmland.

For farmers, that gap is a missed opportunity: solar can create a new, predictable revenue stream, hedge against volatile energy and input costs, and keep land in productive use instead of being sold off. But to even test whether it makes sense on their fields, many are told to pay $10k-$50k and wait 4-12 weeks for a feasibility study, just to get one layout and a rough answer.

Most farmers are not GIS or energy experts and they want to know, in plain terms:
"Can I add solar without sacrificing my best acres, and will it actually pay off?"
TeraWatt exists to answer exactly that question, in minutes instead of months.

What it does

TeraWatt is a plug and plan solar simulator built for farms. A farmer draws their field, sets a budget and goal, and gets a site specific solar layout that respects crops, terrain, and local rules.

It's designed around decisions farmers actually make:

Protect good ground: TeraWatt helps keep prime crop acres in production and pushes solar toward less productive or easier to build areas, instead of blindly filling the field.
See the money, not just the map: Every layout comes with CAPEX, expected annual kWh, and simple payback, so farmers can see whether solar helps stabilize income or risks over spending.
Plan around real constraints: The system bakes in topography, weather, and zoning setbacks, so farmers aren't surprised later by "you can't actually build there."
Compare realistic options: Farmers can flip between "Max Output" and "Cost Optimized" scenarios to see how different choices affect both yield and usable farmland left for crops.
Ask questions the way they talk: With voice first analytics, a farmer can literally ask "What if I only use the back 20% of this field?" or "How much does payback slow down if I lower my budget?" and hear a clear answer tied to the map.

The outcome isn't just a pretty visualization; it's a farmer ready answer to "Is solar worth it here, given how I use this land?"

How we built it

The architecture consists of:

Digital Twin (Google Earth Engine + NASA POWER API + RAG): Fetches elevation, irradiance, and zoning data to build site specific field models that respect farmer operations.

Simulator (React): 3D canvas where crops and solar arrays coexist. Renders fields as farmers see them (blocks, rows, access paths) instead of abstract formulas.

Energy Architect (Heuristic optimization): Turns farmer goals like "max kWh under 50k" into feasible crop + solar layouts respecting terrain, setbacks, and equipment access.

Voice Layer (ElevenLabs): "Interview your land." Converts spoken questions into scenario changes with audio explanations synced to the visual layout.

Backend (Next.js): Coordinates data pipelines, agent execution, and real-time simulation updates across the full stack.

Challenges we ran into

Translating data into farmer language
Elevation grids and irradiance curves don’t say “this low spot always stays wet” or “this ridge pays off faster.” We had to design layers that translate technical features into plain language insights farmers already intuitively use.
Layouts that respect how fields are actually worked
Early layouts would slice fields into weird shapes or block off access lanes. They were mathematically “good” but operationally terrible. We iterated constraints so the Energy Architect keeps access paths and avoids breaking up workable blocks.
Avoiding “AI that argues with the farmer”
If the model suggests panels on land a farmer knows is their best soil, trust is gone. We had to make it easy to lock regions for crops, then rerun layouts that respect those boundaries so the tool feels like a collaborator, not a dictator.

Accomplishments that we’re proud of

From sketch to plan in one sitting
A farmer can go from drawing their field to seeing a complete hybrid crop + solar layout with payback estimates in a single sitting, without calling a consultant.
Farmer‑controllable layouts, not black boxes
The Energy Architect can respond to farmer preferences—“leave this block in crops,” “don’t touch the north ridge”—and still produce valid designs, instead of forcing one “optimal” answer.
Voice that explains tradeoffs in human terms
With voice, TeraWatt can explain why it moved panels off a cloudy corner, or how much revenue changes if only the back of the field is used, in plain language mapped directly to the map.

What we learned

Farmers care more about tradeoffs than raw output
A layout that slightly reduces kWh but protects good crop land and keeps operations simple is often more attractive than the “max output” configuration.
Trust comes from alignment, not just accuracy
The tool gains trust when it behaves like a knowledgeable farm advisor—acknowledging “this is your best ground” and offering alternatives—rather than just pushing pure engineering optima.
Voice is best when it answers “why should I trust this?”
The most useful voice interactions were “explain this choice” and “show me a safer option,” not generic Q&A about solar.

What’s next for TeraWatt

Income‑focused scenarios
Templates like “maximize stable side income,” “minimize payback period,” and “keep premium soil untouched” so farmers can pick scenarios that match their actual goals.
Deeper policy and program awareness
Integrate local incentives, co‑op programs, and PPA options so layouts come with context like “this design qualifies you for X credit or program.”
Farmer‑ready export packs
Generate one‑pager summaries, maps, and data slices a farmer can send directly to installers, lenders, or co‑ops to start real project conversations.
More regions and crop types
Expand to more geographies and farming systems, so TeraWatt stays grounded in how different crops and regions actually operate.

Built With

React – 2D simulator and UI tuned for field layouts.
Next.js – Backend APIs and orchestration logic.
TypeScript – Shared types and logic across the stack.
Google Earth Engine – Topography and elevation data.
NASA POWER API – Weather and solar irradiance.
RAG pipeline – Local zoning and setback constraints.
ElevenLabs – Voice‑first analytics for “interviewing your land.”
Custom /data pipelines – Topo + weather feature engineering for farm‑relevant signals.