AI Enters the Age of Reason : “The End of Easy Billions, the Beginning of Real Value Creation”

AI Enters the Age of Reason : “The End of Easy Billions, the Beginning of Real Value Creation”
3 min read

OpenAI’s latest funding round illustrates a shift in tone in the sector: investors are now demanding a sustainable business model and measurable value from artificial intelligence, whereas technological promise was enough until now. According to the published analysis, this turning point could be favorable for the future of AI.

OpenAI raised $11 billion in an initial funding close, while its initial ambition was for an amount that could reach $30 billion. Its valuation stands at around $1,250 billion, below the initially envisaged targets. These figures remain considerable, but it is now the gap between ambitions and reality that is drawing attention.

The company that helped trigger the global generative AI revolution is finding that technological enthusiasm is no longer enough to guarantee unlimited access to capital. Should this be seen as the beginning of a bubble bursting? The author does not think so “necessarily.” He sees it more as the transition from a promise economy to a proof economy, a shift that does not concern OpenAI alone.

Technological Progress, Adoption, and Value Creation

Since the arrival of ChatGPT, the industry has been advancing at an exceptional speed. Models are gaining performance, multimodal capabilities are progressing, and agents can use tools, produce code, and execute complex sequences of actions. These advances are undeniable, but a confusion has set in: it is considered that the increase in technical capabilities mechanically leads to proportional value creation.

A smarter model is not necessarily more profitable, and a widely adopted technology is not necessarily economically sustainable. The number of users, the volume of tokens consumed (units of text processed by a model), or the number of agents deployed are activity indicators. On their own, they do not demonstrate value creation.

An organization can multiply experiments and observe individual productivity gains without a notable improvement in its operational results. The time saved by a developer does not automatically accelerate deployment to production, and an agent that automates ten tasks does not necessarily create value if the overall process remains unchanged. As the author summarizes: “We have learned to measure what AI produces. It is now time to measure what it transforms.”

The Economic Paradox of AI

On one hand, the cost of using many inference capabilities (running a trained model) is decreasing thanks to hardware, software, and architectural optimizations. On the other hand, the race for cutting-edge models requires significant investments in infrastructure, energy, data centers, and computing capacity. Each unit of processing becomes cheaper, but the spending required to remain competitive increases.

Investments also do not guarantee lasting differentiation. Competing models are progressing, open alternatives are gaining performance, and companies are learning to orchestrate multiple models according to their needs and costs. Where will the value lie—in the models, the infrastructure, the platforms, or the transformation of business processes? The author believes it will gradually shift towards this last dimension, without making the value of the other layers disappear.

For user companies, the need to measure results comes with a requirement for flexibility. Their strategy must not depend entirely on a single supplier, and they must retain the ability to change models, control their data, and track their costs. The first phase of the AI revolution was about making this technology accessible to all; the second will have to demonstrate that it can create lasting economic value.

S’abonner
Notification pour
guest

0 Commentaires
Le plus ancien
Le plus récent Le plus populaire
0
Nous aimerions avoir votre avis, veuillez laisser un commentaire.x