Dear Reader,
On April 12, 1996, a promising young tech company went public…
Shares were priced originally at $13. But the stock opened on its first day at $24.50. And it finished up on the day by 154%.
The enthusiasm for shares of Yahoo! were palpable. And it was easy to see why. At the time, the Yahoo! website was the default homepage for many people’s internet browser. And it was the clear frontrunner in the promising new arena of “web search.”
And for a time, it looked like Yahoo! would remain the king. By 2000, the company was valued at $125 billion, up from just $896 million the day of its IPO.
But we all know how that story ended…
In the wake of the Dot Com bust, Yahoo’s valuation would fall to $10 billion. In the years that followed, it would lose market share to a smaller competitor…Google. By 2016, Yahoo! was bought by Verizon for just $5 billion. That’s a 96% loss in value from its peak.
The point of this story is to show us that the “obvious” winners in a new field don’t always stay on top. That’s important context as we navigate the AI megatrend in the years ahead.
Long Live the King?
ChatGPT is the clear front runner in the generative AI space with an estimated 100 million users. Bard, Google’s attempt at AI, has 30 million users.
OpenAI – the company behind ChatGPT – is the king. But can it retain that title?
Last week, Elon Musk launched his latest venture, xAI, along with former employees of Google, OpenAI, and Microsoft.
It’s the latest in a series of well funded AI products to launch since ChatGPT burst onto the scene last November.
It remains to be seen if Musk or any of the other competitors can dethrone ChatGPT in the years ahead. But what’s important to know is that ChatGPT no longer has this field to itself. And it’s much too early to assume it will be the “king of AI” over the long haul.
After all, it wouldn’t be the first time a tech company with a head start lost out to a newer entrant. Yahoo! vs. Google was one example. But there were others.
BlackBerry dominated the smartphone landscape in the early 2000s. The company once held over 50% of the U.S. smartphone market.
The iPhone launched in 2007 and by 2013 it held 40% of the market share. Blackberry was down to just 5.9% of the market.
Since the iPhone’s launch, Apple’s shareholders have seen a 5,300% return. Meanwhile, Blackberry’s shares have fallen by 91%.
Today, most investors are solely focused on which chatbot is going to be the winner – ChatGPT, Bard, or a newcomer like xAI. If history is anything to go by, it’s too early to declare a winner.
But in the battle for AI supremacy, there are other ways to profit.
That’s because all of these competing AI’s all require immense amounts of hardware. Regardless of who wins out… hardware makers will profit.
Picks and Shovels
Until 2012, the demand for computing power doubled every 24 months. This doubling cycle is down to just two months thanks to cloud technology and AI.
This has already created a 50% increase in demand for Nvidia’s chips. CEO Jensen Huang raised its revenue forecast. In effect, it stuffed two years worth of growth into a single quarter. Shares are up 214% this year.
Microsoft and OpenAI have reportedly bought hundreds of millions of dollars worth of Nvidia’s chips to build ChatGPT.
Musk reportedly bought 10,000 Nvidia GPUs for his xAI project. At an estimated cost of $10,000 per A100 chip, that’s $100 million in hardware. He even quipped, “seems like everyone and their dog is buying GPUs at this point.”
But It’s not just going to be the chip makers that benefit. AI servers are large clusters of processing chips connected by optical cables. In 2023 Google estimated less than 5% of the entire AI servers cost were the optical components
As the demand for AI computing power expands it will create a larger market for these components.
AI servers will also need far more memory than traditional data center servers. Sanjay Mehrotra the CEO of Micron said Chat GPT servers require five times more memory than a standard server.
As more AI servers are built, the entire hardware industry will benefit no matter who wins. The VanEck Semiconductor ETF (SMH) is up 55% this year. On July 14, it reached a new all-time high.
That’s why I’m not even thinking about picking a winner out of ChatGPT, Bard, xAi or any other.
What is certain is that these AIs will require growing amounts of computing power to stay competitive. For now, the chip and component makers are the real winners.
I’ve already shared one of my favorite AI hardware companies, AMD, in a previous letter. You can read more about it here.
Regards,
Colin Tedards
Editor, The Bleeding Edge