Will AI Co-Pilots Deliver on the Hype?

How can we turn the power of artificial intelligence to fighting climate change, and will the recent advances in generative AI deliver impactful applications specific to climate? These are the questions that drove Buoyant's deep dive into climate AI over the past four months, led by MBA fellow Shivali Rao. Given our focus on digital climate solutions and the fact that most of Buoyant's portfolio companies use AI in some core way, we wanted to update our thesis to consider recent AI advances. We concluded that AI is a useful tool but that the need to keep experts in the loop was more important than ever especially related to material science discovery and industrial energy applications.

The introduction of Generative AI has brought us to the top of the AI hype cycle (Gartner), prompting Buoyant to take a critical look at this technology and evaluate its true impact. Data showed us that venture capitalists poured a staggering $23.2B into Generative AI startups in 2023 and that the Gen AI market will reach $1.3T by 2032. Some emerging use cases have significant climate impacts, and our summer research focused on two areas: climate co-pilots and materials generation.

‍

Figure1: Artificial Intelligence Hype Cycle

‍

Exploring AI Co-Pilots

The term "co-pilot" has been popularized and become synonymous with Generative AI, especially in the business setting (Microsoft's Generative AI tool is even called 'Copilot'; other incumbents such as Salesforce and SAP also have deployed their versions of AI co-pilots).

Keeping climate in mind, we define co-pilots as tools that augment (rather than replace) specific roles or functions in key climate industries such as energy, agriculture, materials, and manufacturing. Our research uncovered, however, that rather than being climate-focused, many startups are deploying industry-agnostic co-pilot solutions that are focused on delivering business value over climate benefits (Aisera, Skypoint, and Crux, for example, provide co-pilots for a range of business functions such as sales, finance, and marketing). Instead of focusing on these broad applications, Buoyant homed in on four domain areas that have climate impacts: Business, Research, Operations, and Industry.

• ‍AI for Business: AI copilots that increase productivity in key climate industries (energy, agriculture, food, etc.) or for key climate roles (data analysts, regulatory professionals, sustainable investment professionals, etc.)
• ‍AI for Research: AI copilots that improve experimentation and R&D accuracy in scientific domains such as chemistry, physics and biology.
• ‍AI for Operations: AI copilots that optimize operational tasks for key climate workforces (utilities, renewable energy, and other field work).
• ‍AI for Industry: AI copilots that automate industrial processes to reduce energy usage, reduce waste outputs, and support sustainable manufacturing processes.

‍

‍Figure2: Co-Pilot Market Map:

With this lens, we noticed that many promising startups evaluated in our research do not position themselves as "Generative AI" companies. Instead, they are energy data companies, materials informatics companies, recycling companies, etc. that incorporate Generative AI features into their software.

For example, HData, a Buoyant portfolio company in the AI for Business category, is a data analysis platform with a Generative AI feature that enables energy professionals to be more productive and accelerate the energy transition by dramatically reducing the time involved in mining regulatory filings for data for development projects, rate cases, and legal briefs. Citrine Informatics, a startup in the AI for Research category, is a materials informatics platform that augments a scientist's role, allowing them to perform experiments more efficiently and accurately, thereby accelerating the time to market for sustainable products.

Delving into Materials Generative AI

Research into co-pilots led us to dive into the Materials Science industry and evaluate decarbonization efforts in the space. Calls to address the industrial sector’s high emissions have drawn the attention of companies trying to decarbonize the industry, many of whom are leveraging generative AI technology. One promising use case is applying Gen AI to aid in the development of sustainable materials and products; some estimate this market to grow to $3.2B market by 2032.

Both startups and incumbents are operating in this space and have deployed models to discover and design sustainable materials with varying levels of success. Google, for example, announced that they discovered 2.2 million new molecules using their GNoME platform. However, reviews of their findings showed that few, if any, novel, credible, or usable materials were discovered, questioning the technology's ability to make strides in the materials science industry.

Interestingly, our research into the use of technology to develop sustainable materials uncovered a slew of startups going to market without AI. Enerpoly, for example, developed a Zinc-Ion battery, a sustainable alternative to Lithium-Ion batteries which are currently dominating the renewable battery market, without a proprietary AI platform informing its design. This suggests that with the right signal, brute force experimentation can generate sustainable alternatives in the materials science industry (see below market map for additional examples).

More value lies with the material than the discovery tools

Another key learning in our research is that given startups' ability to develop materials without AI-aided design and the need for sustainable materials to be produced at scale to decarbonize the industry, the true value of a material is its production and ability to commercialize over the discovery of said material. This takeaway is reinforced in the biotech industry where Generative AI has been used to discover new drugs, therapies, and proteins. In one example, a startup used AI to discover a novel drug inhibitor and proceeded to bring the products to market. In the subsequent acquisition, the Generative AI component contributed to ~2.5% of the value of the drug inhibitor, suggesting that the inhibitor itself held a greater value rather than its discovery.

These findings lead us to segment the sustainable materials development market based on startups’ use of AI technology:

• Tech Agnostic: Startups focused on bringing a sustainable product to market that do not use technology to help in its design or discovery
• Tech Enabled: Startups focused on bringing a sustainable product to market, using technology as a key input / differentiator to their offering
• ‍Tech Driven: Startups focused on developing proprietary software to design / discover sustainable molecules, compounds, or products

‍

Figure3: Sustainable Materials Market Map

While Tech Agnostic and Tech Enabled businesses are more concerned with bringing sustainable materials or products to market, Tech Driven businesses are more focused on improving the R&D function by expediting the experimentation process and improving the accuracy of the discovery of novel sustainable materials.

It's also unlikely that many Tech Enabled businesses will focus on generating numerous materials rather than focusing on the manufacturing and commercialization of a few newly discovered materials that has demonstrated value. This insight, coupled with the fact that the materials industry operates with low margins making it difficult to charge royalty fees, lead us to conclude that Tech Enabled or Tech Driven businesses that are asset-light, have a licensed product design they can manufacture through joint partnerships, or SaaS-based business models that provide access to proprietary software and datasets, are most attractive in this market (like Citrine).

Conclusion

While Generative AI has value and can make a meaningful impact on climate, it is more likely a valuable feature rather than a defensible stand-alone product. A common denominator amongst strong Gen AI solutions is its ability to increase the productivity, efficiency, and accuracy of a person or a system. That value in the context of climate superpowers critical workforces and enables the accelerated decarbonization of key industries such as energy, materials, or manufacturing. At Buoyant, we are excited by climate focused startups that are able to tap into the true value of Generative AI - we are actively digging into these solutions so if you are a solution provider or startup in the space, please reach out!

Buoyant hosted a webinar with a panel of experts where we presented our research and facilitated a discussion with a number of start-up CEOs. You can view highlights from that webinar below.