Friday 29 Mar 2024
By
main news image

This article first appeared in The Edge Malaysia Weekly on February 27, 2023 - March 5, 2023

UNLESS you have been hiding under a rock or living in a deep dark cave without internet access for the past 12 weeks, you have probably heard far too much about artificial intelligence (AI)-powered ChatGPT. The natural language chatbot has been wowing the tech world with its ability to do everything from composing college essays to writing complex computer code. Although a handful of art-related AI programs such as DALL-E 2 had begun to proliferate on Instagram and other image-oriented sites around the same time, it was ChatGPT, created by the start-up OpenAI and backed by Microsoft, that caught most people’s imagination and investors’ attention.

Because a lot has been written about it, and there is so much hype around it, I will try to stick to the basics. Generative AI aims to understand and predict information from a particular data set. It is an old tool that has only recently started to ramp up. You may have probably come across it as it was previously used in applications like email, as Smart Compose, which allows the email program to finish sentences started by a user.

Essentially, ChatGPT uses AI to produce human-like responses to search queries, based on data scooped up from the internet. Ask it to solve a complex math question and it will do it in seconds. You can also get instant translation into English of text in a foreign language like French, German or Portuguese, for that matter. You can ask it to write a story, a poem or lyrics for your next song. However, while a ChatGPT-written story may not have any spelling mistakes and the lyrics it writes might rhyme, if you read carefully you might find that some, or most, of it just doesn’t make sense. You can even ask it to pick the best high-growth stocks. Just don’t be surprised if it recommends a bunch of pump-and-dump penny stocks no one has heard of. Or you can give it an inventory of what’s in your refrigerator and your kitchen cabinet and it will promptly spill out recipes that make use of only those contents.

What makes generative AI special is just how quickly it has gone viral. In its first two months, ChatGPT attracted over 100 million users. UBS’ internet analyst Lloyd Walmsley notes that it took viral short-form video platform TikTok about nine months from launch to reach 100 million users, while Meta Platforms Inc’s Instagram took two years and six months to get that number. “We cannot remember an app scaling at this pace,” Walmsley says.

Little wonder, then, that Wall Street has been comparing the arrival of generative AI that creates content to Apple Inc’s launch of the iPhone just before the start of the global financial crisis 15 years ago. With supply chain concerns, persistent labour shortages and high inflation in the aftermath of the pandemic shutdowns, the world is looking for new products and services that will help lead it out of the current slowdown to the next phase of growth in 2024 and beyond. Venture capitalists and tech investors who have borne the brunt of the new higher-interest-rates-for-longer environment see a glimmer of hope in ChatGPT and other generative AI services. Chinese tech players such as search giant Baidu Inc and e-commerce giant Alibaba Group Holding Ltd are readying their own generative AI-based search applications.

US$1 trillion market?

When Wall Street gets too excited about a technology, it starts building a business case for it. Walmsley says that he has heard venture-capital investors speculate that the market for generative AI applications could be as large as US$1 trillion (RM4.4 trillion). He notes that the world has over one billion knowledge workers and OpenAI charges US$42 a month for the professional version of ChatGPT. Assuming every one of those knowledge workers, whether they work from home or in an office, gets two accounts — one general and one specialised — it adds up to not far from US$1 trillion.

In reality, there aren’t a billion people on earth willing to pay US$42 annually for any­thing, let alone a souped-up search engine that pretends to be slightly smarter than Google. What’s becoming clear is that generative AI is being built up as the saviour of the global economy, just as the iPhone was after the 2008 financial crisis. Wall Street is always looking for the next new thing. Yet the iPhone wasn’t just another cellular phone. It was a handheld computer and a communications device that allowed the user to surf the net, play music, watch videos, send and read emails, take photos, store documents, play video games and make phone calls on the go like other cellular phones. Groundbreaking as it is, ChatGPT just can’t be compared to the iPhone or other smartphones that came along in its wake, just yet.

The way I see it, the generative AI boom is becoming more of a stock market bubble. Over the past few weeks, a number of AI-­related stocks have doubled, tripled, then corrected a bit, only to bounce back a few days later. Take C3.ai Inc, a firm that develops software for Enterprise AI applications. Its ticker symbol on Nasdaq is AI. The company listed on Nasdaq in December 2020 at US$42 a share. On the first day, the stock rose to over US$100. Within weeks, it had risen to US$183, or more than quadruple its initial public offering price. But by last December, the stock had fallen to US$10.16, or down 93% from its peak two years earlier. Then suddenly, ChatGPT arrived and created a new narrative for AI stocks. By Feb 6, C3.ai stock had tripled from its lows to US$30.92 on high volume.

That’s what happens when we are in the midst of a market bubble. Investors just race to buy anything that has a connection with AI. And what better stock than the one whose ticker symbol is AI? C3.ai stock plunged 30% last week. It has since rebounded a bit to around US$23 with a market capitalisation of around US$2.5 billion, but it is still a slow-growing software company that helps industrial firms use AI. It is unclear how the boom in generative AI would help grow its revenues and put it on the path to profitability.

Another AI pure play is BigBear.ai Holdings Inc, 20-year-old firm that develops operational AI software used in cyber engineering and business agility. It has a market cap of just US$480 million but has notched several government contracts for AI in recent years. BigBear stock soared 740% from early January to Feb 6. It has fallen 44% since, though it is still up 375% over the past eight weeks. A third purely AI stock is SoundHound AI Inc (market cap US$800 million) whose shares soared 260% in the first five weeks of this year. SoundHound, which facilitates customer service phone calls and conversational AI that replaces human interaction, has seen its stock pare back a bit since but is still up 210% from the start of this year.

In 1999 and 2000, as the tech correspondent for a Hong Kong-based regional news magazine, I had a front row seat to the bubble in tech and dotcom stocks. It was a challenging time to be a tech journalist and remain focused on the big picture. Long before the bubble burst in March 2000, my colleagues and I were betting on how soon and how badly it would all end.

Clearly, the market’s recent fascination with all things remotely linked to AI has turned into a bubble. Having lived through plenty of bubbles — real estate, biotech, crypto as well as the dotcom bubble, to name a few — I can tell you that one thing all bubbles have in common is that they eventually burst. The other thing is they tend to last slightly longer than we all think they might. Unfortunately, that’s when many a rational person loses his mind.

How to play the boom

When you find yourself in the midst of the transformation that we are about to witness in the AI space, the best course is to stay away from hyped-up beneficiaries whose prices have already soared skywards.

During the gold rush, the people who made more money were not those digging for gold or those who owned the land that was being dug but the makers of the picks and shovels that were used to do all the digging.

So, how should investors play the ChatGPT/AI boom?

For AI, those picks and shovels can be software or hardware firms or, for that matter, a combination of hardware and software bundled within a sophisticated chip. Nvidia Corp is the world’s leading producer of high performance graphics processing units (GPUs), which are key to AI, cloud computing, virtual reality, automation and blockchain — all leading tech sectors. Nvidia has a market cap of over US$550 billion, more than five times that of Intel Inc, which until four years ago was the most valuable chipmaker on earth. Another player, Advanced Micro Devices Inc, or AMD, which competes head-to-head with Nvidia for chips used in gaming as well as data centres for cloud computing, is trying to carve a niche for itself in AI-centric chips as well. I wrote about Nvidia in this column in February 2018 (“Can Nvidia remain a chip powerhouse?”) when its stock was at a split-adjusted US$58. Despite a roller coaster ride, it has almost quadrupled since. The stock reached an all-time high of US$333.41 in end-November 2021 but started to slip as cryptocurrency prices plunged. Nvidia chips are used in mining Bitcoins and other crypto coins.

Bank of America’s chip analyst Vivek Arya notes that Nvidia has what it takes to lead the generative AI “arms race”. For one thing, he says, the company has a “full stack of accelerated silicon, systems, software and developers”. Arya believes Nvidia’s sales and earnings could see compound annual growth rates of 25% to 34% over the next five years “as generative AI adoption increases, which in turn will help dramatically grow the total addressable market of the giant chipmakers’ accelerator business to US$62 billion by 2027”.

So how big is the generative AI market? Deploying ChatGPT to run Google search would require over three million GPUs and cost US$80 billion or as much as what ­Amazon, Google and Microsoft — the three big cloud infrastructure providers — spent on data centres in 2021, says Pierre Ferragu, an analyst at New Street Research. Nvidia’s top-of-the-line GPU chip sells for over US$1,600. Even if Nvidia sells the chip for US$1,000, allowing for huge volume discounts, that would be US$3 billion in additional revenue. Add in Microsoft’s Bing and others that will be spending on chips for generative AI and you can see it makes a substantive difference to Nvidia’s bottom line.

Over the next six to 12 months, there will be a new rush of companies trying to rebrand themselves as AI companies or AI-centric firms. Tens of billions of venture capitalist money will be poured into AI start-ups. That’s what the market wants and that’s what venture capitalists and investment banks are going to give investors. Here is my two cents’ worth: Only a handful of companies will thrive and make their way into the Big Tech league. The rest will fade away as soon as the AI mania subsides. And even those that make it will have plenty of scary moments. Remember the dotcom bubble. Take Amazon.com, the one company that made it through way back during the bubble. Just an online bookseller at the time, its stock peaked at US$113, or a split-adjusted US$5.65, in March 2000 before collapsing 96% over the next year or so. The stock has risen nearly 20-fold since, though it is still down 49% from its all-time high in July 2021. Whatever you do, just tread carefully in this AI bubble.

 

Assif Shameen is a technology writer based in North America

 

Save by subscribing to us for your print and/or digital copy.

P/S: The Edge is also available on Apple's AppStore and Androids' Google Play.

      Print
      Text Size
      Share