The AI goldrush and the nuggets left behind
06 juillet 2023
It seems like it happened overnight. Artificial Intelligence (AI) went from something distant that would not typically come up in small talk between colleagues (unless you work in the field, maybe), to the next great game changer that promises to transform the way everything is done. ChatGPT is now already a household name that can write homework for students, find bugs in programmers’ code or even draft a basic contract according to your specification. Just two months after the release of ChatGPT by OpenAI in November 2022, it had already become the fastest-growing software application in history with well over 100 million users. Subsequently this led to massive investments by Google, Baidu, Meta and many others to develop competing technologies to maintain their market shares.
Every time a new technology or concept is the subject of such interest, the media spends an incredible amount of resources theorizing all the different ways in which the world will be made different by this new shiny thing. Yet very few of these promised changes end up making a lasting impact. Blockchain was supposed to revolutionize the world of finance and decentralize the entire system, yet it remains a niche technology with a damaged reputation from the crypto craze.
The metaverse was sold as something that would be part of our everyday life with most households having their own VR headset, much like everyone had their own PC. We’re not quite there yet, although Apple now seems to believe this is the next big thing.
Finally, 5G was meant to unlock so many opportunities with promised better connectivity, low latency and no pocket loss. Every company was mentioning some opportunities from the internet of things to cloud gaming. All these frenzies had something in common: actors that overpromise and underdeliver in spectacular fashion, at least in the short term.
Then comes the new buzzword: generative AI. Between the Q4 2022 and Q1 2023 earnings season, the mention of AI in earning calls more than doubled among S&P500 companies. Nvidia rallied almost 200% in 2023 as one of the most obvious potential beneficiaries and is now worth over $1 trillion while trading at 40x revenues. For reference, these are dotcom bubble multiples.
Artificial intelligence as a field of research goes back to the late 1950s, 15 years after the famous Turing test was first put forward. So how did it become all the rage overnight over 60 years later? Long story short, significant advances in Large Language Models (LLM) in the last decade paved the way to commercial use of generative AI. In a context where the global economy has needed new productivity catalysts to prevent GDP deceleration and to help reduce inflationary pressure in the post-COVID era, ChatGPT seems to be filling the void almost too perfectly. As a new productivity tool that could potentially impact almost every aspect of the service economy, it’s no wonder everyone is jumping on the hype train. Nonetheless, if the commercialization of the internet is any reference, and assuming the impact of AI is of a similar size, we are still at the beginning of the dotcom/AI bubble where euphoria is high, and every company is set to either benefit from AI or disappear into irrelevance. Seasoned investors who lived through the internet bubble may offer one or more of the following pieces of advice:
- Your internet/AI stock pick will likely prove wrong in the long term, as it will take years to fully realize the impact of the technology, while regulations can hinder companies’ plans.
- Your stock pick could be the right one, yet the current valuation may still not make sense.
- There are solid companies outside the tech/AI bubble that are being unfairly penalized as investors sell them to buy into the new hype stock.
Within this speculative environment, we identified an opportunity to initiate a position in a company that had been on our radar for a long time: Keywords Studios (KWS LN). Based in Ireland, Keywords is the dominant player in the fragmented market of video game outsourcing. With studios in over 26 countries, across eight different lines of business and three development divisions, Keywords operates at an unmatched scale three times larger than its closest competitor, yet with a market share of only 6%. The company offers services covering a wide range of developer requirements, including audio services, customer support for live games, marketing and social media management and bug testing.
Keywords had long been a darling in the video game small-cap space, commanding a valuation that made it challenging for us to justify an investment, despite its strong niche positioning and business model. However, near the end of April, the company appeared on an AI loser basket built by Bank of America, based on the belief that most of Keywords’ services would eventually be brought back in-house by game developers due the reduced need for labour caused by new AI technologies. This triggered a downward spiral in the share price. From that point on, negative momentum continued to feed on itself driven by index weight adjustments, loss cutting, quant signals and so on.
On closer examination, it became clear to us that the story was being misunderstood and that investors were selling services companies like Keywords indiscriminately. In fact, Keywords had already been making acquisitions and investments in AI technologies for at least a year before ChatGPT became a household name. It was already using this technology to enhance its localization services (Kantan AI), customer support business line (Helpshift), and to improve on its quality testing expertise (Mighty Games), among other things. Furthermore, Keywords is uniquely positioned to benefit from exposing its machine learning systems to a variety of games, languages and codes. It has a scale advantage that individual video game developers cannot match.
So why did we choose to invest in Keywords and not a game developer that owns its own intellectual property (IP)? The global video game market is highly hit-driven, which introduces risks and revenue lumpiness for developers, especially in the small-cap space where the number of IPs a company holds is usually limited and few games are released each year. Additionally, there is a significant ramp-up time when a new project is undertaken, as the game developer’s workload is not consistently aligned with that of the audio or functional testing teams, meaning employees may not always have the necessary workload to keep them on payroll.
In this environment, it is easy to understand why video game companies of all sizes are increasingly turning to outsourcing various stages of development. This is where Keywords excels. By working with virtually all the top gaming companies in the world, the company can leverage its scale to provide a consistent workload to its studios. Furthermore, by working across an unparalleled variety of games, Keywords builds a unique breadth of expertise without the need to manage its own IP or take on the risks associated with the release of a single title. Keywords is a great way of betting on the growth of the video game industry without making a call on specific titles or the medium on which it is consumed. The company represents most of what we would look for in a core portfolio holding: a leader in a niche market with pricing power, strong secular tailwinds and a good track record. And we got to buy it at a discount to its average valuation, thanks to investors that fell for this new AI mania.