Terranova: AI-guided land lifting as a low-cost alternative to conventional flood defences

The problem

Parts of San Rafael, a city of around 60,000 people just north of San Francisco, are sinking. According to NASA, Bay Area cities, including San Rafael, have been gradually sinking at a rate of more than 0.4 inches per year over the past 50 years, as layers of mud, sand, and clay compress under the weight of the buildings and infrastructure built above them. Altogether, the city has already sunk by 3 to 4 feet.

At the same time, the water is rising. According to a NASA-led study, sea levels along the California coast are projected to increase by 6 to 14.5 inches above the late-twentieth-century average by 2050. In the Bay Area, the rise could be even steeper — over 17 inches within the next 25 years. The threat is not theoretical. Between 1950 and 2016, San Rafael experienced 19 flooding events severe enough to be declared federal or state disasters.

According to Chris Cogo, an environmental justice specialist on the city's Sea Level Rise Collaborative Project (SLRCP) steering committee, the neighbourhood most at risk is the Canal District — home to more than 10,000 predominantly immigrant residents and sitting at the lowest point along the bay shoreline.

The conventional response to this kind of threat is to build barriers — seawalls, levees, pumping systems. But the cost is staggering. According to the Center for Climate Integrity (CCI), defending against rising seas could cost communities in the US $416 billion over 20 years — seawall spending alone approaching the scale of the original investment in the interstate highway system. Closer to home, the San Francisco Bay Conservation and Development Commission (BCDC) estimates that adapting to sea level rise will require $110 billion, with over $230 billion in potential damage if nothing is done.

The question is not whether these areas need protection. It is whether there is an approach affordable enough to actually be deployed.

The solution

Terranova is a robotic flood prevention company that takes a different approach to flood defence. Rather than building barriers to hold water back, it raises the land itself — restoring ground that has gradually sunk over decades.

The approach works by injecting a slurry — a thick, pumpable mixture — made primarily from waste wood such as tree trimmings and other organic material, into the ground at depths of around 40 to 60 feet. The material is inexpensive and widely available. Once buried deep enough that no oxygen can reach it, the wood does not decompose, which means the carbon stored inside it stays locked in the ground rather than being released into the atmosphere. Co-founder and CEO Laurence Allen put it simply: it is "like blowing up a balloon under a piece of paper. The surface — and everything on it — rises."

The injection is carried out by an autonomous robotic unit called the Atlas 3, which moves across a work site drilling wells and delivering material, with human operators on site throughout. What determines where and how much material to inject is a software layer built by Terranova. The company has mapped California's underground geology using publicly available data and approximately 700,000 core samples — small sections of earth collected during drilling, most taken during the construction of water wells across the state. A machine learning model analyses those samples to identify which areas face the greatest flood risk. According to Allen, 40 years of geographical projections fed into the system showed that the entire downtown and Canal area of San Rafael would flood if no action was taken. From there, a second piece of software works out the best injection plan for each site by rapidly testing and refining thousands of possible approaches. Once a plan is ready, city planners and contractors can review it using Terranova's own planning tool — an interactive visual interface where they can see and adjust the proposed land elevation before any physical work begins.

While conventional flood defences require large-scale construction projects costing hundreds of millions of dollars, Terranova's approach uses low-cost materials, automated delivery, and software-guided planning to significantly reduce costs.

The impact

1. A cost structure that changes which cities can act

Terranova has quoted San Rafael $92 million to raise 240 acres by four feet, with annual maintenance costs of approximately $1.5 million per year — a figure Terranova believes could be halved if Marin Sanitary, the local waste collector, gathers and transports wood chips to the sites over a 10-year period. To put that in context, Allen says that flood consultants have quoted the same city between $500 million and $900 million for seawall protection.

For a city of 60,000 with limited fiscal capacity, the difference between a $900 million project and a $92 million one is the difference between a project that can realistically be funded and one that cannot. If the cost model holds at scale, it could open flood adaptation to communities that have until now been priced out of conventional infrastructure approaches.

2. A dual-purpose intervention: flood protection and carbon revenue

Because the injected wood waste remains sealed from oxygen underground, the carbon it contains is permanently stored rather than released through decomposition in landfills or incineration. Terranova sees an additional opportunity here: by permanently burying organic waste, the company could apply for carbon sequestration credits — essentially certificates confirming that a verified amount of carbon dioxide has been removed — which can be sold to companies looking to offset their emissions. That revenue, in turn, could help bring down the cost of the project for cities, reducing the bill to city taxpayers.

This is what makes the model distinctive. Traditional flood infrastructure — seawalls, levees, pumping stations — protects a city but generates no additional revenue. Terranova's approach, if the carbon credits are validated, would produce a revenue stream from the same physical work, which could offset costs for local governments.

3. Early traction and interest from beyond San Rafael

The company raised $7 million in its first major round of investor funding in November 2025, according to Berkeley Engineering, with more than three times as many investors wanting to participate than there was space for.

Other cities are already paying attention. Cesar Zepeda, vice mayor of neighbouring Richmond, California, shared that his city is interested in "using local resources, local people and locally developed engineering to protect low-lying areas right away instead of waiting decades to see any protection from more complex and vulnerable levees and pumping systems."

According to Berkeley Engineering, Terranova is actively looking for new projects starting in early 2026 — spanning flood prevention, wetland restoration, and land elevation work across and beyond the Bay Area. Because the underground model the company has built already covers the whole of California, and the primary material is waste wood — available almost everywhere — the approach is not tied to one location. Any area with suitable ground conditions and a sinking problem could, in principle, be a candidate.

Lessons learned

• New technology needs to answer practical infrastructure questions early. San Rafael's climate adaptation planner, Kate Hagermann, raised a concern that any city would recognise: if you raise the ground beneath a building, how does it connect to the street, the train tracks, the highway? The San Rafael experience suggests that for public servants evaluating unfamiliar approaches, cost savings alone may not be sufficient — understanding how a new intervention fits with existing roads, utilities, and infrastructure is an important part of the assessment process.

• Community urgency and planning timelines do not always align. Steering committee member Rita Mazariegos, a Canal District resident, explained how her neighbourhood "has flooded already once or twice, and people lose their things — they have lost cars," and that it would be too early to rule out any proposal. At the same time, the city wanted to complete its feasibility study before committing to a direction. Thorough assessment protects public money, but the San Rafael case illustrates that residents most exposed to flood risk may also feel the cost of delay most acutely. Mazariegos noted that "people don't know where to go because there is no specific plan from the city" — a reminder that uncertainty itself can be its own burden for affected communities.

As San Rafael's feasibility study continues, the city's experience offers an early look at the questions other coastal cities are likely to face — about cost, technology, community trust, and how quickly decisions can realistically be made.

This case study was written with assistance from artificial intelligence.