AI 2026: Strengthened Governance in the Face of Digital Growth

AI in 2026: Between Promises and Challenges

The year 2026 is shaping up to be a decisive turning point for artificial intelligence (AI), with increasingly intense discussions about its impacts. As organizations continue to explore the possibilities offered by AI, uncertainty remains regarding its real effects. Questions about a potential AI bubble, the emergence of artificial general intelligence (AGI), and productivity gains remain unanswered.

First and foremost, it is expected that AI will cause more harm in 2026 than it did in 2025. This does not mean that it will not also bring positive effects, but it seems inevitable that increased reliance on AI will lead to more frequent negative impacts. Tragedies related to interactions with AI have been reported, such as vulnerable individuals going missing or accidents involving autonomous vehicles. While accidents are also caused by human drivers, the omnipresence of AI in our lives ensures an increase in impacts, often positive, but also negative.

We are witnessing a transition from a continuous flow of what is called "AI mush" – synthetic and unreliable content – to a true avalanche. AI-generated articles are now said to have surpassed those written by humans on the Internet. An AI application, Suno, produces the equivalent of an entire Spotify catalog of music every two weeks. This proliferation of low-quality content erodes trust and increases the risks of misinformation, both in society and within organizations.

Incidents of deepfakes have already affected the business world, such as a fake voice of a CFO requesting a money transfer to a fraudulent account. Are companies ready to face AI agents designed for blackmail or malware? I believe that 2026 will be the year when leaders begin to give AI governance the attention it deserves. However, it is feared that painful lessons will need to be learned first before risk management related to AI becomes a priority.

The situation is complex. Many technologies can be used for malicious purposes. A recent open letter to lawmakers highlighted that there are fewer regulations on AI systems that could pose catastrophic risks than on establishments like sandwich shops or hair salons.

I do not share the apocalyptic visions of AI detractors. On the contrary, I see immense potential for benefits to the world. Take, for example, AlphaFold, an AI system developed by DeepMind, which has revolutionized biology by turning protein folding into a solvable challenge. By predicting the structure of hundreds of millions of proteins, it has accelerated drug discovery and opened new avenues in disease research. This is AI at its best: not replacing human intelligence, but augmenting it to unlock previously unimaginable advancements.

A Pragmatic AI

While examples like AlphaFold represent breakthrough innovations, much of the attention on AI in 2026 will focus on more down-to-earth applications. Strengthened AI governance will go hand in hand with a focus on creating tangible business value, what I call "pragmatic AI."

I foresee significant advancements in industrial AI, as it is already quite mature. We have decades of experience in AI-based predictive maintenance, and these solutions continue to create value. Today, we are applying these lessons to various fields, such as engineering and design, energy management, quality assurance, and supply chain orchestration.

I anticipate a portfolio approach: industrial players will experiment with cutting-edge models like GPT-5 or Gemini Ultra while recognizing the opportunity offered by smaller, specialized language models. Chinese open-source models like Qwen, DeepSeek, ERNIE, or Wu Dao are designed for efficiency, delivering good performance with reduced computational intensity.

Generative AI interfaces represent another significant innovation for the industry, allowing engineers to "converse" directly with the infrastructures they oversee. I expect growing interest in these user experiences based on natural language.

In 2026, companies will shift from cost-optimized computing supply to geographically secure infrastructures, in order to regain strategic control over AI usage. Deloitte estimates that nearly $100 billion will be invested next year in sovereign AI computing capabilities. A clear trend towards what Gartner calls "geopatriation" is emerging, used as a strategic hedge against volatility. This represents a profound shift from the priority given in recent years solely to cost control, recognizing that complexity and risk are here to stay.