Google: Bard Isn't A Failure

Last month, we saw Google (NASDAQ:GOOG) release its Bard chatbot to much fanfare and commentary. The early response was muted, with many reviewers saying it performed less well than ChatGPT and Bing at many tasks. Later, some reviewers noted that Bard was faster than Bing, but the overall impression from most reviews was that ChatGPT gave better quality replies.

This mixed response to Bard might be part of why GOOG has underperformed Microsoft (MSFT) this year. The underperformance has not been massive (about 4%), but it was significant earlier in the year, when MSFT was dazzling the world with its ChatGPT and Dall-E implementations. Microsoft did not just build a ChatGPT box into its browser, it extended GPT’s functionality in several ways, for example by giving it internet access, which allowed it to deal with current events. This made Bing’s Chatbot more useful for some use cases than ChatGPT, but also led to issues like Bing threatening users. Eventually, Microsoft had to make some tweaks to address the issues.

It’s certainly the case that Bard has a ways to go until it catches up with ChatGPT. Speed advantages notwithstanding, its output isn’t wowing early reviewers. However, it would be a mistake to think that is due to a lack of technical capacity on Google’s part. AI applications have to be trained in order to produce output, and training on copyrighted content can create legal issues. The exact nature of copyright ownership with respect to AI is a bit of a grey area, because the field is so new. However, there have been jurisdictions that have ruled that AI art can’t be copyrighted, and some are now saying that training on copyrighted data without consent is infringement.

Viewed in this light, Google’s strategy may be correct. Google appears to be playing it safe, not only with its AI rollout, but with its training data as well. A Search Engine Journal article said that Google uses a mix of C4 data, Wikipedia, websites and forums–with forums making up the largest share of the total. This is a limited set of training data, but it lets Google ensure that it is training its LLM on data that it’s allowed to use.

The slow rollout of Bard compared to Bing also helps with risk management. Currently, Google is rolling out Bard gradually, on a country by country basis. Right now, only the U.S. and the UK are part of the beta test. By taking things one region at a time, Google can get a look at issues (such as copyright claims) as they arise, and learn what the risks are before deploying worldwide. By contrast, Microsoft is already letting people in 169 countries use the Bing Chatbot. Google is playing it safe, and it might win out with this strategy in the end. In this article I’ll explain why I think Google may yet win the “AI wars” and why I think the stock is a good risk/reward opportunity today.

Why Slow and Steady Might Win The Race

When comparing Google and Microsoft’s chatbot strategies, we can see that Google’s is much more conservative. Google was slower to launch its chatbot. It was slower to demo its LLM functionality. It was slower to talk to the media about Bard. It has been far slower to roll out its chatbot to users in multiple countries. Basically, Google is being more cautious than Microsoft is.

You might think that this is a bad thing for Google. In a tech industry that popularized mantras like “Move Fast and Break Things,” being slow to adapt may seem like a losing strategy. Indeed, it could become one, but it needn’t necessarily become one.

Apart from the legal issues already touched on, you need to understand two things:

1. LLMs are very expensive to run. Morgan Stanley (MS) once estimated that Google rolling out a ChatBot would reduce EBIT by $6 billion, even if the average response was only 50 words long.

2. Google already has an overwhelming majority of global search market share. If it can hold its share without launching a chatbot, then it can save a lot of money, because chatbots mean extra costs.

Basically, Google is right to play it safe with Bard, because it may be more profitable without it. Microsoft CEO Satya Nadella recently talked at length about how the “Gross Margin of Search would drop” due to chatbot expenses and increasing competitive pressure. The comment was a dig at Google because, while these higher expenses would impact Bing as well as Google, only the latter has search as a large percentage of its revenue mix.

Nadella’s comments accurately describe one possibility. But a question arises:

What if Google can retain its market share without a chatbot? What if all this extra spending fails to increase Bing’s market share?

If that turns out to be the case then Google can keep its moat without any incremental spending on AI. That would be a huge win for GOOG. And there have been some early indications that this scenario will in fact play out. For the month of March, StatCounter had Google Search at an 83.17% worldwide share, down just 0.2% month over month. GOOG’s share actually increased in February, when Microsoft’s chatbot rollout was underway. From the end of December to today, Google’s search market share increased! Put simply, Bing isn’t cutting into Google’s action in any major way, so it may make sense for Google to keep its Chatbot fairly low key. It may be able to retain those sumptuous chatbot-free search margins after all.

Google Valuation

Having looked at Google’s risks and opportunities in AI, we can turn to its valuation. In most of my recent coverage of GOOG, I touted its valuation as a selling point, but now the stock isn’t as cheap as it once was. So, let’s take a look at how Google is being valued today.

We can start with the multiples. At today’s prices, GOOG trades at:

• 22.85 times adjusted earnings.

• 22.76 times GAAP earnings.

• 4.79 time sales.

• 5.2 times book value.

• 14.5 times operating cash flow.

• 22.65 times free cash flow.

Going by multiples, Google stock is not as cheap as it was the last time I wrote about it. However, it isn’t extremely expensive by the standards of big tech either. According to Siblis Research, the NASDAQ 100 has a 23.79 weighted average P/E ratio, so GOOG is cheaper than its index.

Another approach we could use to value GOOG would be to use a discounted cash flow model. This is where you try to estimate how much the company will earn in free cash flow (“FCF”), and discount it at the opportunity cost.

Google’s trailing 12 month FCF is $4.59 according to Seeking Alpha Quant. The most basic DCF model would be to simply assume that GOOG’s FCF never grows in the future (in other words, grows at 0%), and discount it using the treasury yield as the opportunity cost. At the current treasury yield, with no risk premium, Google’s FCF is worth $131. So, there’s a nice amount of upside.

We can also model scenarios in which GOOG achieves some FCF growth–here it becomes more important to use a risk premium because the risk of failing to hit the projected growth is significant. A few scenarios we could model for, and the fair value in those scenarios, are:

• 10% growth for five years, 0% growth after that, 1.5% risk premium: $142 fair value.

• 20% growth for five years, 0% growth after that, 5% risk premium: $120.

• 20% growth for five years, 5% perpetual growth after that, 6% risk premium: $199.

As you can see, all of these scenarios end up with Google having fair value that exceeds its current stock price. Basically, GOOG just needs to achieve 0% growth to be worth more than the price it trades for. In other words, Google has to stop earnings from declining, as happened last quarter. A bit of cost cutting could be all it takes for Google to be worth the investment at today’s prices. If Google can maintain its search margins without having to do an enormous chatbot launch, then it may well end up achieving positive earnings growth–plus, even 0% growth justifies an investment today.

The Bottom Line

The bottom line on Google Bard is that it’s no failure. Rather, it’s an approach to chatbot development that is suitable for Google. Cautious, methodical and prudent, it helps test the waters with mass LLM deployment without risking the company’s juicy search margins.

To be sure, there are some risks here. If Bing manages to take some market share away from Google search, then GOOG will have to play catch up. On a whole year basis, that hasn’t happened, although the latest data from March does show Bing with a very small share gain. Overall, Google’s market position is solid, and it continues dominating its main product category. That should translate into shareholder value eventually.