AI can process 15 million molecules in a day, yet it still hasn't cured Alzheimer's

In April 2026, The Next Web (TNW) published an article questioning the hype around AI in pharmaceuticals. The technology does speed up individual stages of drug development, but the drug discovery revolution is, in the outlet's assessment, radically overstated. Health chatbots pose a documented danger, and the most severe diseases, including Alzheimer's, remain unsolved despite impressive computing figures.

A telling example: 15 million molecules for one disease

The key example TNW cites is the work of the pharmaceutical company Novartis. In late 2025, a team of researchers working on Huntington's disease used generative AI to computationally design 15 million potential compounds of a particular type of molecule. That is an enormous volume, several orders of magnitude more than a traditional lab could work through in a comparable time by hand, or even with classical high-throughput screening.

It is precisely such figures, millions of AI-generated candidate molecules, that are usually offered as proof of a revolution in pharmaceuticals. TNW does not deny that generative models can explore chemical space at a scale no human can. The problem, in the outlet's view, lies elsewhere: the scale of candidate generation is not the same as the scale of real clinical success. Between "designing 15 million molecules" and "getting a working drug that has passed clinical trials" lies a long road on which the vast majority of candidates drop out.

Why the hardest diseases remain unsolved

The main thesis of the TNW piece: where real breakthroughs are needed, in neurodegenerative diseases such as Alzheimer's, in Huntington's disease and in other complex pathologies that are poorly understood at the molecular level, AI has not yet turned the situation around. The reason is not a lack of computing power but the fact that the biological mechanism of such diseases is itself poorly understood. A model can generate millions of plausible candidate molecules, but if science does not fully know which target must be hit to halt the progression of Alzheimer's, even the most powerful generative system will be optimizing the wrong problem.

Key points of the TNW piece:

• The AI revolution in drug discovery is real but, in TNW's assessment, radically overstated
• Medical health chatbots are a documented source of risk, not only of convenience
• The most severe diseases remain unsolved despite AI progress
• Example: in late 2025, a Novartis team generated 15 million potential compounds for Huntington's disease therapy

It is telling that TNW chose as its example not a startup promising a revolution but a large, conservative pharmaceutical company with a long history: Novartis. This is an important detail. It is not a marketing claim from a small AI company that needs to attract investment, but a working process inside an organization that by definition treats bold promises in drug development with caution and is obliged to confirm any result through long and expensive clinical trials.

What this means for the biotech industry

The TNW piece works as a counterweight to the wave of enthusiastic headlines about "AI that will solve cancer and Alzheimer's within a few years." In practice, generative AI in pharmaceuticals today copes best with the part of the process that lends itself to formalization: generating and initially screening chemical structures, predicting their properties, and optimizing already known molecular scaffolds. But where a fundamental understanding of a disease's biology is required, as with Alzheimer's and other neurodegenerative disorders, computing power alone cannot replace decades of missing scientific knowledge.

For investors and pharma companies, this means assessing more soberly which stages of drug development AI genuinely accelerates and which remain as long and uncertain as they were before generative models, and not mistaking impressive numbers of generated compounds for a guarantee of clinical success.

The second part of TNW's thesis also deserves separate attention: health chatbots as a documented source of risk. Unlike laboratory use of AI, where a faulty candidate is simply weeded out at the next testing stage, an error by a chatbot giving medical advice directly to a user can lead to real harm without any intermediate oversight from a specialist. In the outlet's view, this difference in risk level deserves far more public attention than it currently gets.