ChatGPT, Google Square off in Tech Battle of Titans
If there was ever a company which looked dominant and unstoppable a mere few months ago, it was Alphabet. Their Google search engine commands a 90%-plus share in most of the markets it operates in. Then along came ChatGPT.
ChatGPT has, deservedly, sent shockwaves around the world. I am not mincing my words here — it will change the internet as we know it. The internet has literally just changed from a repository to an AI engine.
Chat GPT is the fastest-growing consumer product in history. It had over a million users in its first week and more than 100 million users after just two months. Previous technology juggernauts haven’t come close: TikTok took nine months to get to 100 million users, Instagram took nearly three years and even Google took nearly two years to reach this milestone.
It isn’t just the rapid growth of users of the platform that’s interesting. It’s that it demands a review of the search process as we know it, how we perform searches literally and the resulting business model which underlies it. It may even redirect us away from advertising and the prevailing surveillance capitalism model.
At the moment it looks like Open AI, the developers behind ChatGPT, have everything to gain but behind the scenes is giant tech overlord Microsoft. If all goes to plan they could be the unexpected winner in AI, and there are literally trillions of dollars in market capitalisation at stake. Microsoft’s 23 January $10 billion deal with Open AI may well be the tech deal of the century.
It is ironic, given Alphabet has been at the forefront of AI development for well over a decade and their core business is essentially also a large language AI. But it does seem like this is the first time Google’s search supremacy has been threatened since it took the lead from Yahoo way back in 2002.
Although Microsoft stumbled after antitrust action in the late 90s and essentially ‘missed’ search, mobile and social, it still has a huge market capitalisation of just under $US2 trillion. If Microsoft manages to escape the boundaries of its ‘Office Suite’ of products, it could become the world’s biggest company.
The Genius in the Deal
The genius of the deal Microsoft made with Open AI is in “cloud credits”. To understand why, we need to know that the biggest cost in developing and operating an AI model is the incredible amount of computing power required. And Microsoft is the second-largest provider of cloud computing globally with Azure. Microsoft not only gave Open AI large funding across to tranches of $US1 billion in 2019 and $US10 billion in 2023 for an equity stake, but it did so knowing that Open AI needed cloud computing, which they included as part of the deal – in cloud credits. Essentially, Microsoft was giving money to Open AI to literally pay back to Microsoft. They also gained a vital equity stake in the most important technology launch in the past decade. As a part of the most recent deal, Microsoft will have an exclusive access to Open AI’s product suite, and will gain a 49 per cent share of Open AI. However, Open AI will need to give back Microsoft 75 per cent of the profits until Microsoft recoups its initial investment.
So, why would Open AI sign what seems like such a one-sided deal? Well, beside the fact that the founders will make very large sums on the equity stakes, Open AI is burning through vast amounts of cash. Currently, Open AI is reported as spending more than $3 million on cloud services per day. As their usage grows, that number is expected to grow and result in at least $3 billion spend on cloud computing this year. Before Open AI did its first deal with Microsoft, it paid Google Cloud, which would eventually become a key competitor, $120 million in 2019.
The Next Web
The intention of Microsoft and OpenAI is to build the foundations of the AI era in the hope that it becomes a platform others work with and on top of. In real terms ChatGPT is really just one application of the AI model GPT 3.5 and each subsequent iteration of GPT which stands for Generative Pretrained Transformer. This is part of the shift I’ve written about here on the Eureka Report, where we go from searching for what we need, to creating what we need.
It’s important to realise that GPT isn’t just a consumer facing service. It can and will be plugged into many other firms’ consumer operating systems and intranets as a back end service. In addition to this, Microsoft hasn’t wasted any time integrating into their other suite of products. It already is available through their Bing search engine and its edge browser. A key difference is that ChatGPT in Bing is using the live web to gain its data, and is not limited to data stopping at 2021. Microsoft also has plans to integrate it into MS teams. But it won’t all be smooth sailing. All AI models to date include bias, misinformation, and can even provide dangerous results where recommendations are given without sources or verification. This has been especially evident in the live Bing version. Then there is the cost equation.
Even if Microsoft is currently being paid by Open AI, the cost per ‘prompt’ is currently around $0.02c. This is vastly more than the $0.00001 per Google search, and probably couldn’t support a pay per click or display advertising model. The recent option to subscribing to ChatGPT for $20 per month is a clue as to where the business model of Generative AI is likely to go – The Subscription Industrial Complex. This would both remove the free rider problem, and temptation to compromise product quality to appease the advertising model supporting it. If OpenAI / Microsoft can get 100 million subscribers, and $20 a month, that would result in $24 billion in new revenue. Microsoft also claims that for each market share point of search it gains, it will generate $2 billion in revenue.
Think about it – if we shift our search habits to ask questions and getting an actual answer, rather than a page of links and options – the pay per click model will die alongside it. Bing might just become the world’s first Premium Search engine – a pay to play for a different kind of search. The market is likely to bifurcate into two segments: Search (Traditional web links) and Creation (Generative AI).
The Code Red which was called in through halls of the GooglePlex hasn’t resulted in anything that seems like a worthy response to CHATGPT. After a failed demo last week of the Google AI chatbot Bard, it lost more than $100 billion in market cap. But I also wonder if the market senses that Google has far more to lose even if (and most likely when) it develops a competitive AI product. Fifty eight percent of Alphabet’s revenue comes from search, which is driven by pay per click advertising, which simple can’t survive with generative AI. Microsoft only gains 5 per cent of its revenue from Bing pay per click advertising. In real terms, it has a potential ten-fold search revenue upside, with near zero downside all the while potentially adding a new weapon to its already strong enterprise offers of Windows, Office and Azure.
Just when we thought we thought a one tech firm could never be usurped, a new technology comes along which potentially changes everything.