Microsoft has announced a multi-year, multibillion dollar investment in artificial intelligence (AI) as it extends its partnership with OpenAI the creator of popular image generation tool Dall-E and the chatbot ChatGPT.
ChatGPT is able to provide believably human responses to questions and Microsoft is backing it to revamp its own technology, including its search engine Bing. However, ChatGPT isn’t the only bot in town. It’s just part of a years-old trend towards better and better Natural Language Processing (NLP). NLP forms the crucible within which those looking to assert their AI dominance have sought to burn most brightly. We highlighted this in our recent 5-yr presentation where we showed the evolution of leading-edge NLP platforms, as expressed by nodal density (ie complexity, which approximates to efficacy):
ChatGPT is based on OpenAi’s “GPT” NLP platform. It was launched in 2018 with 150mn parameters, and has increased in complexity over the past 5 years to the point that GPT-3 now has 175bn parameters – so, more than 1100x more parameters today than 4.5 years ago.
ChatGPT uses “GPT3.5”, a tweaked version of GPT-3 with broadly similar capabilities. Microsoft paid OpenAI for exclusive access to commercialise GPT-3 in 3Q 2020. – it’s one of the key reasons that we own a position in MSFT in our fund. ChatGPT’s capabilities are very impressive – they are the best available at the moment, but not by a huge margin. DallE is another AI platform that “imagines” and creates images from text. In the following link, Dall-E was asked to imagine the broader context for a number of famous paintings, and then “paint” the larger canvass in the style of the relevant artist. You’ll see the outcome for Vermeer here:
https://openai.com/blog/dall-e-introducing-outpainting/
Stepping back a little, our core thesis is that we expect AI to permeate the entire economy over time. The inference from this is that we entirely expect MSFT’s products and services to be deeply imbued with AI, and, with OpenAI as a key partner, driving GPT across all of MSFT’s services is consistent with this view.
Will it impact Google's search business?
How this relates to Google’s search business is a more nebulous issue. Google has their own leading edge NLP platform, “LaMDA 2”, which is highly capable, and caught the headlines when a Google employee alleged it to be sentient last year. So, GPT3 and ChatGPT have the headlines at the moment, but it is not long since Google’s leading edge NLP was in the zeitgeist: these things tend to leapfrog one another, and so far it has paid to stand back a bit, and try and see the big picture, which is, of course, that AI is getting better and better, and becoming more useful… it’s too soon to declare GPT and MSFT to be the winners.
More prosaically, there are costs to consider. Morgan Stanley recently produced a good piece of work looking at the computing cost to process Google’s search queries on the one hand, and ChatGPT queries on the other: the cost per GOOGL search was about $0.005/search… the cost per ChatGPT3 query was between $0.03 and $0.14. This doesn’t matter when ChatGPT is being used in relatively small scale, but it really matters if ChatGPT were to be used at scale. If ChatGPT were to be deployed at great scale, and if it were to be successful in displacing GOOGL search somewhat, the graphics chip company nVidia could be well placed to benefit in our view.
GOOGL’s advertising business is currently experiencing its first proper cyclical downturn – albeit nothing on the scale of that seen at META or ROKU. It’s pretty easy to throw mud in GOOGL’s direction at the moment and we expect that some of it will stick. With the stock trading at c.12.5x PE, and with huge optionality ahead, it earns its place in our portfolio (Sanlam Global Artificial Intelligence Fund), but not at the weight it enjoyed in previous times.
Important information
The Fund may invest in shares of companies listed on stock exchanges in the United Kingdom, and outside the United Kingdom, exchange rate fluctuations may cause the value of investments to go down as well as up. Investing in companies based in emerging markets may involve additional risks due to greater political, economic, regulatory risks, among other factors.
This document is provided to give an indication of the investment and does not constitute an offer/invitation to sell or buy any securities in any fund managed by us nor a solicitation to purchase securities in any company or investment product. It does not form part of any contract for the sale or purchase of any investment. The information contained in this document is for guidance only and does not constitute financial advice.