Why you should short Google tomorrow

Alphabet Inc (NASDAQ:GOOG), the parent company of Google and Youtube, is literally shitting itself. So much so that CEO Sundar Pichai has dramatically issued a “Code Red” to the engineering teams at Google, to try and figure out a product response to the shocking overnight success of ChatGPT.

It is a perfect example of how a company who dominates a segment of the market to such an extent that it enjoys a near-monopoly can be lulled into a complacent sense of the permanence of its lead while competitors stealthily position to overtake.

And really, now in retrospect, it is astounding that Google should let such an upstart like OpenAI suddenly eclipse in the one area where Alphabet shareholders would have safely assumed an awareness of the threat and a plan in place to deprecate it.

But such was not the case, and Alphabet shares have now reacted by shedding 12 percent of its value since the public launch of ChatGPT a month ago.

Since Google’s Code Red alarm sounding, there suddenly appear news articles from never-before-seen websites touting Google AI tools that are poised to compete with Open AI’s ChatGPT.

Some less-than-imaginative commentators think the threat to Google posed by ChatGPT are overblown, citing Google’s speed and depth of search results, compared to ChatGPT’s famously error-riddled results.

But this vision is rooted in static thinking.

The important thing to consider – and the greatest indicator of the scale of the threat to Google’s ad business (its bread and butter) – is the fact that, despite the obvious deficiencies and relatively dated information set that ChatGPT has to draw from in its current version, the performance and utility of ChatGPT relative to Google’s search results, is orders of magnitude better.

And that’s not even considering the persistent contamination of Google search results with ads and undisclosed placement of search results.

For example, when searching financial news on Google, more often than not, Motley Fool appears as a search result, despite it being strictly a subscriber-driven financial service. There are no journalists, and it is not news. But they do spend prolifically on Google Ads and are among the largest spenders on Google Ads in the financial newsletter space.

Google also sews the seeds of its own destruction by censoring results and rating information sources according to biases that can only be described as political in nature. The disclosure of Google’s routine manual re-organization of search results along political lines is detailed by Google whistleblower, Zach Vorhies, a former Google senior engineer, in his book entitled Google Leaks. 

He writes, “that’s what I believe my former employer was doing with its censoring, algorithms, and other tricks designed to bring about what they believed to be a better world.”

Not that ChatGPT is going to be any better, once it is faced with the conundrum of not just satisfying regulators to keep operating, but satisfying shareholders at some point to keep ad revenue cascading in, assuming that is the business model.

ProPublica, the non-profit journalist-based corporate watchdog says, “Google’s ad business helps fund dangerous disinformation that puts public health and democracy at risk around the world, earns money from millions of gun ads while publicly claiming to block them, and allowed a sanctioned Russian ad tech company to harvest data on potentially millions of people, including possibly those in Ukraine, putting their security and privacy at risk.”

But so far, ChatGPT and OpenAI seem to be focused on deriving revenue solely from user subscription fees, which can already be signed up for here. Pricing is presently incredibly inexpensive, though it is thought that will change with improved performance in future iterations.

Couldn’t Google blow ChatGPT out of the water if it wanted?

It is important to understand that in the world of AI language models, “parameters” are to Chat Bots as gigabytes are to computer processing power; the higher the number, the more powerful the platform. (up to a point). Parameters are pieces of the model learned from historical training data and basically determine the skill of the model on a problem.

OpenAI’s ChatGPT is said to have 100 billion parameter model, and is based on Google’s Transformer architecture.

But the next generation of Transformer is Pathways Language Model, described as, “a 540-billion parameter, densely activated, Transformer language model.”

Transformer , according to the New York Times, is a “neural network architecture that Google Research invented and open-sourced in 2017”

Google’s direct answer to ChatGPT  is called LaMDA, or Language Model for Dialogue Applications. According to former Google engineer Blake Lemoines (who was fired for publicly stating that LaMDA had become sentient), LaMDA is at least 3 times more powerful than ChatGPT.

But Google is hesitant to release its own AI platform for the simple reason that it runs the risk of cannibalizing Google’s ad revenue. But a parallel threat is that Google is worried about the backlash from its users and customers if its own AI platform were to be seen as tainted by racism, gender bias and false information. Or political bias of any flavour.

And herein lies the essence of the threat to Google.

ChatGPT’s next iteration is expected to be released in Q1 2023, and is being touted as featuring “100 trillion parameters”.

If that means the inaccuracies and biases of the current version fo ChatGPT can be largely mitigated, then theoretically ChatGPT would become the default search engine on Earth practically overnight.

Google might then be forced into a position where it must launch LaMDA to compete with ChatGPT, and its ad business would then be under double threat, both from itself and OpenAI. ChatGPT is still free, and without any advertising. If Google is to compete with ChatGPT, any chatbot it releases would therefore have to be free as well.

Regardless, this competition should be the focal point for investors in 2023, since the likely outcome is that Google’s share price is going to plummet, while OpenAI, which is at this point a private company, will see its value absolutely skyrocket, and it will probably be the biggest IPO in history.

Either way, Google/Alphabet is as safe a short as Tesla has been these last weeks.

Original article: Why you should short Google tomorrow

©2022 Midas Letter. All Rights Reserved.