GenAI is increasingly becoming a staple in consumer technology, particularly with the integration of small language models in edge-AI and on-device AI applications.

However, monetising these AI capabilities poses significant challenges, as evidenced by the tepid consumer response to premium AI services and the high operational costs associated with GenAI.

Low adoption of premium consumer GenAI services

The uptake of Google's Gemini Advanced, its AI premium service under Google One's subscriptions, has been underwhelming among consumers, with a subscription rate of only 0.01% of Google One’s 100 million user base despite free trial promotions.

Even OpenAI’s paid users comprise less than 3% of its total user base of 350 million (as of June 2024). This suggests a disconnect between the service's offerings and the perceived value by consumers.

GenAI: staggering operational and environmental costs

OpenAI, despite its pivotal role in the GenAI boom, is not yet profitable and faces a challenging path to revenue generation.

The costs associated with GenAI are so high that OpenAI is losing more money than its making. OpenAI forecasts a loss of approximately $5bn in 2024 due to elevated operational expenses, despite projected revenues of $3.7bn.

The most significant of these costs is the computing power acquired through its partnership with Microsoft. OpenAI projects profitability only by 2029. GenAI requires training a model and then using the trained model to make predictions on new data. Both processes are associated with significant energy and resource consumption.

The International Energy Agency (IEA) has highlighted the growing energy demands of AI, which could double by 2026 to 4% of global energy demands, equating to the entire electricity usage of Japan.

Nuclear power to the rescue?

To address the environmental impact of GenAI, tech giants are investing in alternative but expensive energy sources, such as nuclear power.

Microsoft's September 2024 agreement to revive a unit retired for economic reasons in 2019 at the shuttered Three Mile Island nuclear plant in Pennsylvania in the US to produce energy for Microsoft marks a pioneering move in the tech industry's adoption of nuclear energy.

Google followed suit with its agreement with California-based Kairos Power to purchase power from small nuclear reactors built by Kairos. Amazon has partnered with Dominion Energy Virginia, Energy Northwest, and Talen Energy to investigate innovative approaches to advancing nuclear energy, including the creation of more compact and expedited nuclear reactors.