Will AI drive 10% GDP growth?

Likely no.

An oven in my fictional bakery makes 10 cakes a day. Whatever I can’t sell by the end of day, I have to throw out. If I buy a new oven that can produce 10,000 cakes, do I bake all 10,000?

Likely no.

The intuition is that the bottleneck at my bakery is selling the cakes, not producing them. If I could sell more, why wouldn’t I produce more?

The production arm of AI is either chip makers (Nvidia) or cloud / API infra providers (AWS, Google cloud, groq, FAL) or model producers (Gemini, Open AI) all investing heavily.

Consumption is also rising exponentially in multiple domains like Search (Gemini, Perplexity), AI engineers, lawyers, doctors, personal trainers, accountants, etc.

However, the scale of new consumption generated (consulting an accountant x times a day instead of x times an year) is nowhere near a few $100x bil an year.

With the means of production becoming cheaper, even in areas with explosive consumption growth, the ability of firms to charge a higher value is tricky.

In real terms, if I used to visit my human accountant, doctor, etc. once an year. It is unlikely that I keep paying them as usual and additionally hire an AI accountant for 10% cost. Also unlikely that the accountant uses AI to provide a personalized service for which they charge 10% more.

However, if my accountant used to charge $x per hour and now discount by $y, more people use them and more often.

Also, sectors directly feeding into the supply side like electricity or steel producers will feel a growth pull.

In aggregate, some sectors will likely see explosive growth but the probability of 10% aggregate increase is super low.