We are in the middle of quarter and annual results announcement season and what we see is amazing. Nvidia has posted $6B pure profit over the AI Boom. Microsoft has surpassed any other company and is now with over $3 Trillion the largest company in the world – and Nvidia is 3rd. AMD has announced a strong AI Chip to be on par or better than what Nvidia has to offer. Meta has announced to upgrade its data centers with 350,000 Nvidia H100 cards to improve their AI performance. After already many AI investments, for example in Anthropic, Amazon increases its investment and spends an additional $1 Billion into AI start-ups.
Gen AI continues to be THE main trend and is also fuelling our economies with profits. This means it is not a hype anymore really but has arrived in the mainstream valley. Or has it?
Well, as you know I have been and still are a big fan of Gen AI (you can read about it in my previous posts or hear in my podcasts). But a deeper look at the above first paragraph shows something interesting: What you see there are investments of companies that WANT to make money with selling AI services and profits of companies that deliver the Gen AI PLUMBING or INFRASTRUCTURE. What is absent from this are companies that actually profit from AI usage in their daily operations.
A recent study that was published in the Economist shows a contrasting picture. Only a staggering number of 5% of surveyed US companies used (Gen) AI to produce goods and services and only 7% plan to do so in the coming six months. Usage of (Gen) AI has not yet arrived in the regular economy. This is also supported by the fact that Microsoft only generated 20% of its Azure growth of 2023 through (Gen) AI services. This sounds like test & learn, not like deployment.
If you are looking back, (Gen) AI and its hype only began around 16 months ago, we are still very fresh in this market. And it is for sure a revolutionary technology that changes possibilities and moves barriers. But that also means that firms first need to learn how to actually make the best use of it. For me it is no surprise that we are where we are. It is expected. It wasn’t different when the PC was born, it wasn’t different when the internet started to be a thing. Firms have to first find their way through these endless possibilities that (Gen) AI bring, and employees have to learn to utilize a new tool.
I see another year of test & learn as well as trying to find out which of the Gen AI horses will be the most successful – and what strategy will be the winning formular: One main Gen AI to support throughout the company and across functions and integrates most use-cases (similar to an ERP) or best-of-breed targeted Gen AI solutions for specific use cases which means you would have like hundred different models in your firm. Or will (Gen) AI just be a feature in your already implemented systems landscape? I don’t know the answer, but getting it wrong might mean that you lose momentum and your competitors will gain from you. And so let me be clear, this post is not to make us stop on testing & learning, it just calls out that there is a long way to go.
On the other side, the (Gen) AI companies still need to find the right balance between investments and value. What is the right price for (Gen) AI? This should not be based on what great things it can do or how revolutionary it is or how scarce it is (at least not in the mid- to long-term) – but it must be based on the value it actually can and will create in firms, on fundamentals. How much can you increase productivity? How strong can you reduce costs of R&D or even production? How much quicker can you go to market with a new product? The problem is that we don’t know yet and this means that it is difficult to find the right pricing for (Gen) AI solutions while the service providers are investing the big bucks and must see a return soon.
And this is also what will define where we are in the hype-cycle. In 2024 we must either see the value of (Gen) AI in the real world, in production of Goods and Services – and not in niche, but wider market – or we will see a major crash similar to the dot com crash as all the money that went into infrastucture and model development is not met with revenue or even profit. It will be a tough year for (Gen) AI and for sure a very interesting one to follow.
