Alphabet Stock’s other bet: a deep dive into DeepMind (NASDAQ:GOOG)
Many articles on Seeking Alpha are about Alphabet Inc. (NASDAQ:GOOG, NASDAQ: GOOGL) finance, general affairs and outlook. This is not one of those articles. Instead, it is DeepMind: a division of Alphabet listed in reports alongside Waymo and other business models under “other bets”. I intend to give an overview of what DeepMind already contributing to Alphabet and what potential there is.
AI is probably the most important thing humanity has ever worked on. I think it’s something deeper than electricity or fire.
Sundar Pichai (CEO of Google)
A little history
DeepMind was a British startup founded in 2010 by Demis Hassabis and others. He is still the most famous face in the business today. Early investors included Elon Musk and Peter Thiel. In 2014 DeepMind was acquired by Google (Facebook (FB) was also interested) and since then it has aimed to “solve intelligence”.
Past achievements: from computer games to protein folding
Even before the acquisition by Google, DeepMind was primarily engaged in solving older computer games. The idea was to let an artificial intelligence learn the game from scratch, play against itself and learn from its defeats and its victories. In this way, the natural learning behavior should be imitated as best as possible.
After the acquisition, DeepMind tackled increasingly complex games. For example, in chess, DeepMind’s AlphaZero defeated the best chess engine to date after just 4 hours of training. Engines beat humans at chess for many years, but most were powered by massive databases and calculated other moves with powerful computers. The fact that DeepMind’s AI beats other engines with this independent learning which is an impressive success.
AlphaGo caused even more of a stir when it defeated then-world champion Lee Sedol in a best-of-five Go game. No AI had achieved this before, partly because Go is not computable. The number of possible moves is too high: Go is considered one of the most complex games of all time. There is a documentary about this duel that is worth watching.
Another achievement that has grabbed headlines is called AlphaFold, which predicts 3D models of protein structures. All findings are stored in a freely accessible database.
A system like AlphaFold that can accurately predict protein structure could accelerate progress in many important research areas for society. AlphaFold is already used by our partners. For example, the Drugs for Neglected Diseases Initiative (DNDi) has advanced its research into life-saving cures for diseases that disproportionately affect poorer regions of the world, and the Center for Enzyme Innovation at the University of Portsmouth uses AlphaFold’s predictions to help design faster enzymes to recycle some of our dirtiest single-use plastics.
The long-term goal is the development of a general-purpose AI. However, all experts currently assume that this will not be feasible before 2030. A general-purpose AI would theoretically be able to combine all available information from all fields and apply it creatively, in the same way that humans do. can. For example, it would no longer be possible to determine whether one is talking to a human or a machine when used in a voice AI. Such an AI would probably offer us suggestions for creative solutions to all our problems and render human research useless since it would now be inferior anyway.
DeepMind believes that reinforcement learning is enough to achieve general AI in the long term. Until then, however, AI will be applied to smaller sub-areas to optimize processes and fix problems. Similar to what has already been successfully tested in games.
Is Does DeepMind still make money?
There are two ways DeepMind can contribute to Google’s bottom line. Either to generate revenue or to reduce costs. So far it’s the latter that’s been happening, which usually causes less of a stir than the big revenue numbers, but is just as valuable. DeepMind generates “revenue” by applying its technology to other Alphabet projects. And these have more than tripled (!) to £826m in 2020, according to the latest financial statement. Spending fell from £717m to £780m.
In 2019, the company lost another £477 million. Thus, from 2019 to 2020, sales and profits tripled for the first time. As an independent company, DeepMind would undoubtedly have a valuation in the billions, given these numbers and the intellectual assets it has accumulated.
Of course, you have to be careful with all of this. Since this is revenue within the same company, Alphabet may be artificially inflating this revenue a bit. But it is more likely that DeepMind developments have tangible benefits that increase revenue or reduce costs elsewhere at Alphabet. There are also examples on DeepMind website.
Another observation: DeepMind is one of Alphabet’s “other” bets, but I don’t know how this revenue is accounted for. Indeed, the total revenue from “other bets” in 2020 was $657 million, which is lower than DeepMind Reported Revenues.
Or DeepMind is already operational
Here are some examples of what DeepMind is already doing this right now.
- Identify eye diseases: DeepMind’s algorithm can detect over 50 sight-threatening eye diseases as well as leading experts in the field. Additionally, the system can even predict who is most likely to develop age-related eye diseases.
- Save electricity: Google’s data centers must be constantly cooled, which consumes a lot of energy. DeepMind helped optimize this cooling, saving 30% (!) of power.
- Increase energy production: It was new for me, but Alphabet exploits 700 MW of wind energy. DeepMind has developed algorithms that use past performance and weather forecasts “to make optimal hourly commitments of delivery to the power grid a full day in advance. This is important because energy sources that can be scheduled (i.e. that is, which can provide a set amount of electricity over time) are often more valuable to the grid.” (The source). As a result, the value of the energy generated has increased by approximately 20%.
- More natural-sounding AI voices: WaveNet is used in Google Assistant and other systems to make artificial voices more human-like and improve user experience. Among other things, it mimics natural pauses in breathing.
- Ancient texts supplementing: Archaeologists usually only find fragments of ancient stone tablets and other relics. Bringing them together into a meaningful whole is extremely complex. DeepMind has developed an AI that helps archaeologists classify and complete historical texts.
- Traffic prediction: DeepMind has helped improve estimated time of arrival accuracy by up to 20% to 50% by making better predictions of surrounding traffic.
Many additional opportunities
- Applied to sports, big data can lead to better decisions: Traditional sports is an area where big data has not been used much, even though a lot of money is involved. However, if algorithms can help teams or individual players to make better decisions, it will probably only be a matter of time before they are increasingly used. For example, DeepMind describes a situation during a soccer penalty shootout. An analysis has shown that evaluating various data (past wins, which is the strong foot) leads to much better predictions of where the shooter will shoot: valuable information for the goalkeeper.
- Breast cancer screening: In medicine, the possible applications are numerous. One of them is the more accurate detection of cancer. In one study, it reduced both false positive breast cancer detections and identified cancer cases not detected by humans.
- Game balancing: In the past, it took a lot of work for game developers to create balanced games. A prime example is the real-time strategy genre. Usually, the player can choose several races or countries, and none of them should be stronger than the other. Humans would need thousands of hours to create a balanced system. However, an algorithm could do it in a few hours, and DeepMind is very well positioned here thanks to its experience in the field of gaming.
- Control nuclear fusion plasma: Nuclear fusion could bring a revolution in energy production, but there are still many challenges to be solved.
A control system must coordinate the tokamak’s many magnetic coils and adjust their voltage thousands of times per second to ensure that the plasma never touches the walls of the vessel, causing heat loss and possible damage. To help solve this problem and as part of As part of DeepMind’s mission to advance science, we collaborated with EPFL’s Swiss Plasma Center to develop the first deep reinforcement learning system to autonomously discover how to control these coils and successfully contain the plasma in a tokamak, thus opening up new avenues for advancing nuclear fusion research.
Article: Accelerating fusion science through learned plasma control.
These lists are far from complete. The company’s blog is full of research descriptions, including links that could keep you busy for days. At this point it should be clear that DeepMind has enormous potential.
I think it’s fair to say that DeepMind is still in its infancy. However, many of the previous researches are slowly starting to integrate the social aspects and are becoming financially attractive. The fact that in 2020 alone reported income has tripled is quite remarkable.
In addition, it should also be taken into account that many application domains are only used by Google itself. This means that Google has not yet decided to resell these solutions to other companies. An example of this potential is how sales of Google’s wind turbines increased by 20%. With the many gigawatts of installed capacity and the amount to follow, even minimal optimization would be of enormous value to operators.
As said at the beginning, it is only about DeepMind. Alphabet’s current revenue drivers are, of course, other areas. However, a strong and promising DeepMind is one more reason to continue investing or staying invested in Alphabet.