How to win AI

Chances are you are not the one who is building the next large language model, integrating the next LLM into existing software, or distributing AI in mass. Chances are you are sitting on the sidelines watching, or moderately participating in the products offered.

AI is magic, there is no question about it.

So if you are not the ones who are building or selling this market how do you win.

There are several ways to play the space. But the big two are the productizers and the suppliers.

The productizers

Apple, Amazon, Microsoft, Google, Meta, etc. have announced their AI strategies. They will be incorporating AI into their existing products and services to increase the “magic” their products offer.

They are the ones building AI products. Products like Microsoft’s CoPilot which will come as a $30/month/user add on package to help companies better manage productivity inside of the Office365 suite of products. Microsoft also is working on taking market share away from Google and the search market with Bing and BingAI, integrating that tool directly into the search bar for faster information transfer to consumers. Magic on magic.

Meta is working on building a better advertiser experience with AI and better ad placement and targeting, a direct impact on the company’s top tine revenues.

Amazon wants to help people build better web applications inside of AWS with it’s AI suite of products.

Google is working on a similar product as Microsoft with productivity AI inside of Google workspace and Bard inside of the search engine.

There are going to be a lot of AI products released in the next few years as the space evolves, but not all will win.

Which leads me to the next play.

The suppliers

One of the more interesting places to play in is the suppliers. In order to have a data center, like AWS (Amazon Web Services), process insane amounts of data you need advanced semiconductors.

The suppliers win by supplying the winners with the losers, they supply all, and as demand increases so do revenues.

Graphics processing units (GPUs) offer parallel processing, giving data centers a way to process multiple streams of incoming data at once, and faster than core processing units (CPUs). Nvidia manufactures the best GPUs in the world, so much so that the US government has shut off China from gaining access to these advanced chips, even going as far as banning access to AWS, Azure, and Google Cloud.

Nvidia and AMD are both chip designers. They don’t manufacture chips.

ASML uses extreme ultraviolet light to help make smaller chips, which offer faster compute. It is the only EUV company in the world and some of the machinery sells for $200 million a piece. Every semiconductor company is racing to make smaller and faster chips, and ASML will no doubt capitalize on that.

KLA Corporation is involved in the semiconductor process control market. Semiconductors need incredibly clean manufacturing environments. A single grain of sand can compromise the entire process. KLA offers advanced facility monitoring tools to inform manufacturers about potential problems. More demand means a need for a better supply chain, ideally with less disruptions.

You can even make the case for Intel, Taiwan Semiconductor, or Samsung as the actual chip manufacturers manufacturing the designed chips.

And last one I will talk about is Broadcom. Broadcom currently manufacturers a majority of smartphone chips. Broadcom designs chips that can process incoming information faster. AI isn’t just better processing at data centers, it’s also better Wi-Fi processing and data processing at the individual hardware level.

Supplying the AI boom is a different play in that instead of trying to pick the winning product, the suppliers supply the losers too.

The risks

During the dot com bubble the suppliers, Oracle and Cisco, greatly benefited, so did Amazon. However even the suppliers were damaged by the downturn in the productizers market.

The demand from the productizers harms the supplier market, and everyone loses.

As the losers lose, the suppliers lose this is where the winning product begins to be the right play, but it still may not work as the overhyped market erodes.

There will be money to be made, but at the cost that there will be many who lose greatly.