The economic and ecological challenges of the artificial intelligence revolution

Can its business model stabilize and become profitable? Is its ecological footprint compatible with the energy transition needed to address climate change? Will uses continue to evolve and move away from the search engines we know today?

At a time when this technological shift seems increasingly irreversible, the whole question is whether humanity has the means to adopt this technical revolution in the face of environmental and profitability imperatives.

An economic model under pressure

The craze for generative AI has propelled OpenAI to the ranks of the world's largest tech companies. Yet its entire business model is largely based on expectations of growth and return on investment that seem unlikely to be achieved at the moment.

OpenAI actually spends up to $700.000 per day to run the ChatGPT platform and maintain its operational infrastructure. An astronomical sum that highlights a major difficulty: generative AI is expensive, very expensive.

And forecasts show its losses could triple to $2026 billion by 16. By 2024, when the company will generate about $3,7 billion in revenue, losses could approach $5 billion, with every dollar of revenue costing it $2,35 in expenses.

This imbalance calls into question the viability of the current economic model. 73% of OpenAI's revenue comes from paid subscriptions and the remaining 27% corresponds to revenue from licenses for the use of their LLMs by companies.

This overreliance on subscriptions exposes the company to a major vulnerability: It must successfully convert enough of the free users currently siphoning off its revenue to triple its number of paying subscribers.

The pressure to raise new funds is also becoming an obstacle. The recent $6,6 billion fundraising, valuing the company at $157 billion, is a feat in itself. But how long will investors continue to inject capital without tangible financial returns? With such massive expenditures, is the return on investment achievable in a short enough time horizon to interest the financial markets?

Unlike Alphabet, Meta or Elon Musk with xAI, OpenAI is not as strong and will have to reach even higher levels to finance its growth and satisfy its investors, potentially $200 billion in the next rounds of funding. This situation raises doubts about the ability of a company that suffers such losses to go public, a necessary step to guarantee investor profitability.

A worrying ecological footprint

Beyond the business model, generative AI and the sophistication of language models rely on energy-intensive infrastructures.

OpenAI’s GPT-4 model is estimated to consume up to 3 bottles of water to generate 100 words, illustrating the colossal energy requirements of the data centers needed to run these technologies. American giants must resort to nuclear power plants to support this energy demand. Microsoft has thus reactivated a nuclear power plant, while Google has engaged with the start-up Kairos Power to power its data centers via modular reactors.

While nuclear power can provide stability, its intensive deployment weighs on companies' carbon footprint, as evidenced by Microsoft's 29% increase in CO₂ emissions since 2020. More broadly, these initiatives highlight a broader problem: current infrastructure is not sufficient to meet demand without a substantial energy source.

The International Energy Agency predicts that data center energy demand will double by 2026. This paradox threatens global ecological goals if greener and less costly solutions are not implemented.

At a time when the IPCC is recommending energy sobriety, AI seems to go against climate objectives by requiring unprecedented energy production that is difficult to sustain without a radical transition.

Is it possible to make AI innovation coexist with carbon footprint reduction?

Transformation of uses

The arrival of AI is also transforming search habits: 250 million people use ChatGPT. Users, especially young people, have more immersive and interactive expectations than before. Today, 40% prefer TikTok or Instagram to Google to search for a restaurant, and product searches increasingly start on Amazon (63% of cases).

New AI tools, such as Google Gemini, AI Overviews, Search GPT or Perplexity, offer direct, fast answers and change the user journey while integrating advertising in an innovative way. Although these new uses are still limited compared to the overall number of searches, this diversification could well disrupt the well-oiled economic model of search engines like Google or Bing in the long term. It will be difficult to go back.

Digital marketing professionals must therefore closely monitor these trends to adjust their acquisition strategies and adopt a more multi-channel approach in order to retain and capture the attention of an audience dispersed across multiple platforms.

AI, profitability and ecological transition: a possible convergence?

The development of generative AI faces contradictory imperatives: on the one hand, it requires gigantic investments and exponential energy consumption; on the other, it disrupts usage and monetization models. The profitability objectives of OpenAI and other AI players depend on stable funding and a rapidly growing paying user base, which is difficult to reconcile with ecologically reasonable expansion but also with user habits compared to the free nature of traditional search engines.

AI can, however, be a lever for the development of alternative energy technologies, such as modular nuclear reactors and other less polluting energy sources. In return, these innovations could guarantee the energy supply necessary for the mass deployment of AI. However, the current race for AI requires immediate solutions that risk short-circuiting ecological imperatives. Does humanity have the economic means and the ecological will to integrate AI while avoiding disastrous compromises for the environment?

The future of artificial intelligence: between innovation, profitability and ecological transition

Generative AI is emerging as a revolution with multiple potential benefits, but it relies on a precarious balance between profitability, energy efficiency, and adaptation of uses. As massive investments intensify, the ecological and economic challenges of this technology raise a crucial question: are current infrastructures and resources ready for such expansion? The future of AI could depend on sustainable energy innovations and a redefinition of profitability, failing which the cost of this revolution could well exceed its benefits for society and the environment.

Tribune by Frédéric Jutant, Marketing Manager at Icarus Media Digital (LinkedIn).