Last week Microsoft confirmed the third phase of their long-term partnership with OpenAI, the company that created ChatGPT. They plan to invest as much as $10bn, valuing OpenAI at a staggering $29bn, a big jump from the $14bn they were valued at in 2021.
OpenAI is currently one of the most valuable startups in the US. And, as far as investments go, this is Microsoft's biggest ever. If you add in Microsoft acquisitions, the investment in OpenAI ranks fourth after Activision ($69bn), LinkedIn ($26bn) and Nuance ($20bn).
OpenAI was founded by Sam Altman in 2015 as a non-profit. Original backers for the venture included Elon Musk among others. In 2019, OpenAI established the for-profit arm of the business and Microsoft made their first investment of $1bn. Part of the original deal was that OpenAI used Microsoft's Azure cloud platform to develop and train its AI models and it gave Microsoft first dibs on commercialising OpenAI technology.
Some of you might not have heard of OpenAI but you’ll almost definitely have heard of ChatGPT. Yes, that AI chatbot that can pass MBA exams and produce everything from recipes to poems. OpenAI has developed more than just an AI chatbot though. They also developed DALL-E, an AI system that can create images from descriptions in natural language and Codex, a system that translates natural language to computer code.
The technology behind ChatGPT, generative AI, has the potential to dramatically increase productivity in many different areas. In the near future, existing productivity tools, like Microsoft Office, could automatically provide relevant information, check your work for accuracy and in some cases even do some of the work for you. There are already indications of the sort of productivity improvements we might expect. According to the CEO of GitHub, an internet hosting service owned by Microsoft, around 40% of code produced by developers using their Copilot service is generated by the AI system.
Eric Boyd, head of AI platforms at Microsoft, opined all of this is just the beginning. Interacting with technology using natural language could fundamentally change how we use technology in our daily lives.
The amount of computing power required to train an advanced AI model like ChatGPT is far beyond the reach of most small startups. This means startups might have to partner with a big tech company in order to avoid the large capital expenditures required to build their own AI supercomputing platforms.
Microsoft already offers their AI platforms, Azure Cognitive Services and Azure Machine Learning, to other companies to build and train AI products on. This is a smart move. According to the Microsoft CEO, training models on the Microsoft platform costs around half as much as it would on rival platforms. Even other tech giants, like Meta, use Azure as a strategic cloud provider for its AI ambitions.
But before we get sucked into the media hype we should take a step back. We could be in for a bumpy ride here. People are understandably excited about AI. But they are also worried. It could prove to be a very disruptive technology.
We all like productivity improvements but no one wants to lose their job. History teaches us that disruptive new technologies can temporarily destabilise the labour market. Other areas of society might also be impacted. Reports of students cheating by using ChatGPT has some educators thinking of making students write essays by hand in supervised environments. This is something to keep a close eye on.
Large Language Models like ChatGPT ‘learn’ from a vast amount of data. This doesn’t necessarily make them ‘intelligent’. There’s a saying in statistics about models: garbage in, garbage out. The same applies to AI models like ChatGPT. We mentioned earlier that AI is increasing the productivity of coders by writing some of the code. But is the code written by the AI any good? A recent study by NYU found that around 40% of the code contained security vulnerabilities..
Some highly respected AI experts are critical of the recent hype around ChatGPT. Yann Lecun, head of AI at Meta and winner of the 2018 Turing prize, has been called the godfather of AI. He said that the public’s perception of ChatGPT as an “incredibly new, innovative and unique” tool is wrong. He points out that Google invented the transformer based technology that underlies ChatGPT and have been using it for years in areas like content moderation, recommendations and translations.
In his opinion, big companies like Google and Meta have not pushed harder to introduce consumer facing products like ChatGPT due to ‘major flaws’ like providing inaccurate or even harmful content. Large established companies have a lot more to lose from rogue AI algorithms than small startups. Some of us still remember Microsoft's previous attempt at a chatbot, Tay, and how quickly they had to take it down after it exhibited some concerning behaviour.
Another area to keep a close eye on is regulation. Regulators have been circling big tech companies for years, scrutinising their increasing dominance in the digital world. If only a few big tech companies have the means and capital to train AI models they could become gatekeepers to the infrastructure necessary to deploy AI models. Platforms to train AI models could well evolve into natural monopolies. It makes little sense for every company out there investing in its own AI supercomputer.
Don’t expect other big tech players, like Google, Apple, or Meta, to just let Microsoft’s bold move into AI go unchallenged. Google, given its large share in areas like search, has a lot at stake and they’re unlikely to cede market share without a fight. Microsoft just upped the ante. Expect a response.
Have any questions on any of the stories? Post your questions and we'll share our answers to some of the top questions in our next Market Pulse. Keep in mind we cannot give investment advice - all answers will be perspectives on the public news stories only.
Enjoying these updates? Want to hear more from the Sidekick team as we build the wider product? Sign up to the waitlist here
Sidekick is not yet regulated but has applied to the FCA for authorisation to operate. Prior to Sidekick becoming fully authorised, none of the information provided is intended as an invitation or inducement to apply for any Sidekick product or service.
Please remember, investing should be viewed as longer term. When we launch, your capital will be at risk — the value of investments can go up and down, and you may get back less than you put in.