When the value of just two companies changes by $200 billion in a matter of hours I usually take a closer look. That can even happen when “Married At First Sight”, and not Gaza, has brought you to the point of giving up on humanity. More Gaza later. For now, let’s revisit the events of October 24th. Despite the glow of its recent 25th birthday, Google’s quarterly earnings results failed to impress investors and the subsequent share price dive clipped the guts of $75 billion off the value of the Mountain View tech giant. In contrast, investors were excited by the update on the same night from the world’s second most valuable company, Microsoft, as investors rushed to buy shares and added a cool $125 billion to the valuation of the Seattle tech giant.
The only word on any traders’ lips that evening in New York was ‘cloud’. More specifically, the revenues earned by the critical data storage and processing architectures which support all our personal and business digital apps and services. The ‘cloud’ is where big tech has leveraged its scale and offered enormous computing power to live and work your digital existence. However, these apps and services are now feeding off a new digital super-power – Artificial Intelligence(AI).
Generative AI with its large language models(LLMs) and enormous data learning appetites have turned the cloud into a battle field fought by the big three – Microsoft, Google and Amazon. And, the cloud is flying – not quite literally but Microsoft’s Azure cloud business revenues are rocketing at 29% annual growth rates. Google’s cloud business was perceived the ‘loser’ last week with a growth rate of just….. 22%. You get the picture – the cloud is big money, but it’s also really all about AI. Revenues earned by cloud services (powered by data centres) are a proxy for measuring who is winning the AI ‘war’. Let’s be very clear Google and Microsoft have lots of other revenue channels but there is no doubt that the $200 billion shift in valuations between the two giants was entirely driven by the cloud, and by AI. Still sceptical? Allow me to expand on this thread…
Remember Mistral? Yep, that was the company with 4 guys who raised $120 million with no business and no revenues. Just a PowerPoint presentation. Well, that was 4 months ago. And, now they’ve reportedly raised another $300 million. This time they can actually demonstrate a proprietary large language model(LLM) built with 7 billion parameters for AI training. Yes, built… in 4 months. In valuation terms, Mistral is already a ‘unicorn’ – a startup worth more than $1 billion. If you thought this was merely VC excitement about ‘disruption’ then think again. It feels like the world is still figuring out which of emerging disruptors (with new AI models) or big tech (with its massive proprietary data head start) will win the modelling wars. However, big is still beautiful in investors’ eyes.
Check out all the gloomy headlines – inflation, painful interest rate hikes, war, recession. You’d think stock markets would be cratering. And, you’d almost be correct. If you strip out the share price performance of just 7 technology companies – aka the “Magnificent 7” – then global equities are probably in negative territory for 2023 so far. Now, think about what is driving Apple, Microsoft, Tesla, Google, Facebook, Nvidia, and Amazon who, on AVERAGE, have rocketed in value by 80% this year. For this writer, it is clear these 7 companies possess the best databases on the planet and are in pole position to train AI models to do whatever they want. Some are happy to use 3rd party models like OpenAI’s ChatGPT or Anthropic’s Claude and the investment monies are still flowing fast.
Microsoft has already put $10 billion into OpenAI and the latest reports of funding activity suggest OpenAI’s valuation has jumped from $20 billion to $85 billion….in 8 months. Amazon is putting $4 billion into Claude but, as we have illustrated, there are about 200 billion reasons and counting to be in this race. We can’t forecast the future but it is worth remembering that this is AI in its infancy, or to put it another way, at its worst.
I had the genuine pleasure of chatting to “the Oracle of AI”, Jim Dowling, who presented at an IIBN business event last week. He’s usually based in Sweden and, uniquely, is that country’s only resident lecturer in Deep Learning. It was fascinating to hear him talk about “emerging reasoning” in some of the very large AI models and how lots of well-known businesses are using his company, Hopsworks, to re-configure their data architecture for pending AI applications. What was less fascinating was my estimate that probably 75% of the questions from the audience were fixated on deep fakes, misinformation, AI ‘hallucinations’ and cheating on…. homework. I know, how do we sleep at night!
Now, recall my earlier words that these early building stages are seeing AI ‘at its worst’. Then just repeat one word to yourself, quite a few times. GAZA. As a species we seem to be perfectly good at bringing ourselves to the brink of World War III or demonstrating barbaric behaviours which, on reflection, didn’t quite end with Ghengis Khan or the Inquisition. Bluntly, we can do far better and AI could help – think of education, the unbanked, healthcare, medicine, energy, decarbonisation, urban planning or agriculture. You know, all the bits to do with living. Of course, all important things must have governance and guardrails. How many unapproved foods, drugs or banks do you know? So, get ready for more of the following:
Biden Executive Order Imposes New Rules For AI – ABC News
The excellent Tech Brew newsletter gives a good summary in the following bullets:
The directives in the order cover everything from housing discrimination to bioweapons, and aim to address AI at each stage of development.
Developers must share safety test results with the government, and various agencies will work on developing standards designed to mitigate threats from AI-created biological weapons and deceptive deepfakes.
The order includes a regimen of new privacy research and rules that aims to better govern how developers use information they collect on users.
A section of the order homes in on algorithmic discrimination; it calls for guidance to landlords, federal contractors, and welfare programs on reducing bias in any AI tools they use, as well as new guidelines for the Department of Justice to probe this type of discrimination and more rules around AI’s use in the criminal justice system.
The general consumer protection section focuses mostly on developing standards for AI’s use in healthcare and education.
The order calls for a report on AI’s impact on the workplace, and lists directives for working with allies to implement AI standards internationally.
Meanwhile, over the other side of the pond……
UK, US, EU and China sign declaration of AI’s ‘catastrophic’ danger – The Guardian
Hosted by the British government this week, twenty-eight governments signed up to the so-called Bletchley declaration on the first day of the AI safety summit. One can understand the British government’s eagerness to exhibit some form of responsible stewardship given the stunning revelations coming from the ongoing Covid-19 inquiry in Westminster. An “unfit” Prime Minister surrounded by “f*ckpigs and morons” administering a staggeringly incompetent response to a global pandemic is truly a review for the ages. And a relative reminder of AI’s infancy and humanity’s ability to be……. ehhh…..almost inhuman, or non-human.
So…..GAZA or AI? My money (and clearly a lot of investment capital) is on cloud wars potentially delivering a better humanity. Keep watching, and hoping. It will be worth it.