AI Innovation for Climate Showcased at UC Berkeley Conference
Bakar Climate Labs is a new program, just 3 1⁄2 years old, spun off from the research and incubation program launched under the Biology segment (QB3) of the California Science and Innovation Institutes created at the turn of the century/millenium.
Within that, a core activity is the Bakar Institute of Digital Materials for the Planet or BIDMaP and during April, BIDMap held their first symposium “AI and Climate Symposium: From Cutting-Edge Research to Commercialization” to showcase how the massive advances in large learning models are having a positive and accelerating impact on innovations in chemistry, material sciences and business tools that can deliver a healthier planet sooner rather than later.
The first portion of the symposium offered academic presentations of some astonishing material science research while in the second portion, in keeping with their lab-to-market mission, they presented investors and entrepreneurs leveraging AI for climate businesses right now.
Extreme Tech Challenge was invited to moderate presentations by four great startups assembled by Bakar, XTC and Tenika Versey Walker, Global Head of Sustainable Futures at NVIDIA. Two of the startups had been winners of the 2024 Sustainable Futures Showcase presented as a joint program of NVIDIA’s Inception Program with Extreme Tech Challenge. And we got to present in a prime slot between stellar presentations by Obvious Ventures, Zenskar and Autodesk.
The spectrum was impressive showing impact and deployment RIGHT NOW of solutions for all the climate elements: AIR, WATER, URBAN LAND, MOBILITY and, not least, FUNDING MONEY.
Starting with AIR, we heard from Haoxing Du, Lead ML Engineer at Windborne, one of the XTC/NVIDIA alumni. They are building large atmosphere models for better weather forecasting. That very week, Ms. Du’s colleague was presenting a paper that Ms. Du co-wrote to the 13th International Conference on Learning Representations on “an operational transformer-based global weather forecasting system”
Next, for Urban Land, was our other XTC/NVIDIA Alumnus Jay Sadiq and his company FortyGuard which is building large temperature models to accurately estimate temperature impacts of building location, orientation and landscaping in urban settings.
As interesting as the detailed models both are creating, is to learn how simplistic are our current measurements. Weather balloons just go up on one vertical vector until they burst and then disappear. When you look up “temperature in Berkeley” or “Oakland” you’re just getting one thermometer at the airport or atop an office building. Both Windborne and FortyGuard are building vast data cloud corpuses of readings from which to ‘learn’ far more fine-tuned models of how “climate” works in the atmosphere and at ground level. I also briefly mentioned a 3rd XTC Alumnus who could not attend, 7Analytics of Norway, that is doing similar work with Water and estimating Flood Risk. So, there’s Air, Water, Earth.
For Mobility on Earth, ChargeMate CEO Bradford Crist presented an agentic AI application aimed at providing a one stop UI to navigate and optimize the often nightmarish task of getting an electric vehicle charged. Charging headaches are becoming the leading reason for adoption hesitancy and AI agents are an ideal solution to helping to navigate a new and uncertain topography of charging and thus accelerate adoption of cleaner mobility infrastructure.
Finally, Streamline.Climate presented an amalgamation of 3 separate AI functions to help all of these budding developers spend less time-on, and make less arduous, the process of grantwriting to get support for their climate innovations. Grantwriting can take up to 40% of a researcher’s time. We heard from CTO Douglas Qian how Streamline is both helping the search for the best and most relevant grants, scrubbing the solicitation to discern the key deliverables and trigger phrases, and then helping optimally structure a winning submission.
It was thrilling to engage with these entrepreneurs and a great validation of the core thesis of Extreme Tech Challenge which celebrates how only now in the Twenties are the breakthroughs in semiconductors, microfabrication, robotics, 5G, Cloud, IoT and now the AI wrapped around all of those, able to make affordable high-performance solutions like these. For example, Windborne had a nice diagram showing that current weather forecasts use ballroom sized supercomputers costing millions, while Windborne’s founders started with a handful of NVIDIA-powered game consoles from BestBuy to launch their algorithms.
These are the kinds of practical, deployable, investable applications of extreme technologies that XTC is always looking for, and we sure know how to pick ‘em (with thanks also to our partners as Bakar’s BIDMaP and NVIDIA)
AIR Windborne Haoxing Du
WATER 7Analytics
LAND FortyGuard Jay Sadiq
MOBILITY ChargeMate Bradford Crist