Quality Alone Won't Predict AI Winners

Sony (SONY) had the best tech in town... and nobody cared.

Back in the mid-1970s, long before streaming technology came around, you had two options to catch a show or movie on television.

Of course, you could watch it live... or you could buy a video recorder to tape it. Sony hit the market first with a 1975 video recorder called the Betamax. It was a pretty good machine... The tapes were small, it had good resolution for the time, and it was a quiet machine.

Even so, the technology never caught on. Within a year, a competitor, the VHS, came to market.

Developed by the manufacturer JVC, VHS machines were worse in almost every way. The video cartridges were bulkier. They had worse-quality video and a shorter playback time. And the machines were noisy.

Still, VHS won, easily. It quickly became the dominant format, and VHS machines were soon ubiquitous in American homes.

JVC didn't focus on having the absolute best technology. It focused on something far more important to its success...

Today, I'll discuss the key factor companies need to win a market in the long term and who's focused on it right now in the development of AI.

VHS's one big advantage was its video selection.

Instead of guarding its format the way Sony did, JVC made VHS an open standard and encouraged other manufacturers to adopt it. It was also cheaper to make VHS tapes. That made it easier for movie producers and distributors to start putting their content on video cassettes.

The result was, when consumers went to market, they found a lot more VHS machines and a lot more they could do with the machines.

Instead of focusing on having the best technology possible, JVC focused on having the widest amount of content. The company did a much better job partnering with movie producers and distributors to get all the up-to-date movies and shows.

That way, when folks went to the store to buy one of the machines, they were stuck between the choice of a great machine with very little to play or a machine with content worth watching.

It's the same way Microsoft (MSFT) beat Apple (AAPL) back in the 1980s in the PC market. Apple had better computers... Microsoft made its software easier for developers to use. So again, when consumers were debating what computer to buy... they found they could do more with computers running Microsoft-based software.

Watch for this same phenomenon to play out in AI...

ChatGPT might've been the first artificial-intelligence ("AI") tool available to the average person... but that doesn't mean it's going to be the one everyone talks about forever.

ChatGPT has been around for less than two years. It's only a matter of time before these large language models ("LLM") start commercializing in more ways than a simple subscription.

In fact, Meta Platforms (META) has developed a competing LLM tool called Llama. Even though Llama has made a far smaller splash so far, it could be a frontrunner to overtake ChatGPT.

Llama hasn't commercialized in the same way. It has mostly focused on academic research so far... Meta is mostly working on getting its model right for now. That said, there are many reasons it could pull ahead.

Meta has one of the biggest data sets on Earth... all of it social media data. And even bigger than that, like Microsoft in the 1980s, Llama has chosen to be open source, while ChatGPT does not share its code.

In other words, it's far friendlier to developers than ChatGPT. That means consumers may soon find a lot more things they can do with Llama. While ChatGPT may have the advantage today, we're still extremely early to the AI-model race.

Developers are going to want to work with the company that lets them see the data, and users are going to follow developers... not the first mover.

Wishing you love, joy, and peace,

Joel
August 9, 2024