Bezos Earth Fund is Offering $100M for AI-Led Climate Solutions, Including Alternative Proteins

The Bezos Earth Fund has announced an AI for Climate and Nature Grand Challenge, promising up to $100M in grants for solutions leveraging artificial intelligence – and alternative proteins are one of its key focus areas.

In what is the latest in a line of food-focused investments, the Bezos Earth Fund is committing $100M to harness the power of AI to find solutions that can address climate change and protect biodiversity.

The AI for Climate and Nature Grand Challenge is inviting applications from scientists, researchers, startups, global organisations and non-profits across the world, who are using artificial intelligence to come up with future-friendly innovations.

“AI may have the potential to solve some of our biggest problems, and we’re calling on the planet’s brightest problem solvers to bring their visionary ideas to the table,” said the fund’s vice-chair Lauren Sánchez. “Together, we can innovate and solve these challenges.”

Using AI to make better meat analogues

The competition will run over three years, and its first round will focus on three key areas: sustainable proteins, biodiversity conservation, and power grid optimisation. Subsequent rounds will prioritise other climate categories.

Explaining the premise in a video announcement, Sánchez said: “How can we feed the growing population without hurting the environment? Can AI help sort through millions of protein combinations to produce meat alternatives that are as delicious as beef, but more affordable and with a much smaller environmental footprint?”

The announcement comes a month after the Bezos Earth Fund set aside $60M to establish the Centers for Sustainable Protein, which will address technological hurdles to reduce the production costs, enhance the flavour, and improve the health benefits of alternative proteins. “Alternative proteins are an imperative if we are to stay within planetary boundaries, if we are to feed 10 billion people within those boundaries,” said Andy Jarvis, director of future food at the Bezos Earth Fund. “We’re investing in alternative proteins because they need to be successful.”

This was part of the fund’s $1B commitment towards food systems transformation, with the first investment announced during COP28, earmarking $57M in food-related grants to tackle the threats of climate change and biodiversity loss and preserve food security.

Speaking of which, Sánchez outlined the potential of AI to protect endangered animals and identify species we haven’t discovered. In terms of energy grids, she noted how we could power the world by integrating renewable energy into electricity grids around the world. Outside these focus areas, the fund is also accepting wildcard approaches for “brilliant” solutions that don’t fit the three categories.

“These are complicated, urgent issues, and we believe modern AI can really help,” she said. “But we need to figure out how.”

The Grand Challenge will involve two funding opportunities. In the first phase, the Bezos Earth Fund will award up to 30 seed grants for promising ideas, which will be announced at a joint event with TED during Climate Week NYC (September 22-29). Phase two will see these winners be eligible for grants up to $2M to execute their concepts, in addition to mentorship and support from tech leaders, and access to computing infrastructure and relevant datasets.

While proposals from individuals won’t be considered, the fund says it’s likely that selected proposals will entail alliances between climate and AI organisations. “Can modern AI help counter climate change and nature loss, and, if so, how? That’s the question we hope to answer,” said founder and executive chair Jeff Bezos. “By bringing together brilliant minds across fields, we may be able to invent new ways forward.”

AI’s own climate footprint is a massive issue

“The future is unlikely to be characterised by straight lines and gentle curves, but rather by unexpected changes and tipping points, good or bad,” said Bezos Earth Fund president and CEO Andrew Steer. “The arrival of AI will potentially help solve very difficult challenges.”

But AI’s relationship with climate change is frosty. It’s true that the tech has enabled researchers and businesses to tackle the crisis in numerous ways – it has helped detect floods, wildfires and deforestation in real time, improve climate pattern modelling, conserve water, and recover recyclables. AI can also analyse crop imagery to find pest or disease problems, and perform tasks humans may not be able to, like collecting data from the Arctic when it’s too cold or conducting research in oceans.

It has had an impact in the alternative protein industry too, with the most famous example being Chile’s NotCo, whose patented AI platform, Giuseppe, scours through infinite combinations of plants to come up with the best blend for its plant-based milks, mayo and burgers. Californian startup Climax Foods similarly uses AI to produce plant-based casein proteins for vegan cheese formulations, and Singapore’s Howw Foods makes Hegg, its vegan powdered egg product, with the help of the tech.

All this is great progress for the sustainable food industry and climate change mitigation – heck, AI can help businesses reduce emissions too. However, AI itself has a massive climate footprint. “We seem to be hearing all the time that AI can save the planet, but we shouldn’t be believing this hype,” said Michael Khoo, climate disinformation program director at Friends of the Earth.

The organisation was the co-author of a report that warns the tech will lead to a rise in water and energy use from data centres, and proliferate climate misinformation. Data centres, which run 24/7 and mostly on fossil fuels, account for 2.5-3.7% of global GHG emissions, and are on track to account for 14% of all emissions by 2040. And even if the AI industry improves data centre energy efficiency by 10%, the demand and usage of AI would likely double the number of these facilities, increasing global carbon emissions by 80%.

Then there’s the development of AI chatbots in the first place. Training ChatGPT used as much energy as 120 American homes over a year, while training the more advanced GPT-4 model consumed 40 times more energy. And running these large language models collectively consumes large amounts of water and energy – Google’s chairman has said that each new AI search query needs 10 times the energy costs of a traditional Google search, while the International Energy Agency predicts that energy use from AI data centres will double in the next two years.

All this prompted Charlie Cray, senior strategist at Greenpeace USA, to label AI as “one of the greatest emerging climate threat multipliers”. So while the Bezos Earth Fund’s investment into AI solutions may be well-intentioned, it needs to contend with the tech’s own climate footprint.