I’m going to have to up my game. Not just tennis. As a frequenter of the occasional business discussion panel, this week threw up a very different type of panelist. The Dublin Tech Summit at the RDS hosted a panel discussion on AI which featured contributions from a meta-human avatar created by AI, named Anja. Quite unnerving in a way. If it had been a horse on the panel, I don’t think it would have unsettled me more. Mind you, the no-clothes Emperor Taco Trump guy can’t be far away from appointing a horse to the Senate soon. Anyway, I digress as humans do. Back to AI, and I was thinking it would be no harm to highlight a few significant AI datapoints and developments which have caught my eye in recent weeks. First, the data.
The Stargate data centre project backed by OpenAI, Japan’s Softbank, Oracle and Nvidia and to be built in the UAE is estimated to eventually have the capacity to consume 5GW of power, For context, that’s the power consumption equivalent of the entire island of Ireland. And, Ireland would already be considered a global leader in terms of data centre capacity as a proportion of the total energy grid, about 21%. Clearly, AI and its critical data centre/cloud infrastructure is moving at pace to meet expected future AI usage demand. The pulse-take on AI investment pace has been chip-maker, Nvidia, who reported quarterly results this week. Revenues for Nvidia (despite Trump China tariffs/blocks) are still growing at almost 70% but this doesn’t quite capture the scale of growth. Two years ago, at the time of ChatGPT’s launch, quarterly revenues at Nvidia were $6 billion. Now, they are at $44 billion. Furthermore, Nvidia plans to invest $500 billion to build AI infrastructure in the US. Note, things have also moved on from ChatGPT and other Gen AI tools (like Gemini and Claude) as the drivers of AI investment. The big move now is to “Agentic AI” or “AI Agency”.
Agentic AI is not a pilot or learning model wanting users to test its knowledge. No, this is the real “doing” stuff which companies are now paying to integrate in their work flows. According to CB Insights research, enterprise AI and copilots will generate $13 billion of revenues by the end of 2025 across a variety of activities from sales to coding to customer service. That’s a growth rate of 155% year-on-year and a wake-up call for most companies; the reality is that their competitors are likely deploying AI to dramatically improve productivity and costs. One wouldn’t want to be in the spectator seats for too long and it’s not just a corporate caution. At a sovereign level, Dubai has offered all its citizens free access to the premium ChatGPT Plus service which normally costs $20 per month. The digital information race is truly ‘on’ but there’s also a hardware story emerging.
OpenAI has just acquired Jony Ive’s AI hardware start-up, Io Products. The former Apple key man, whose design credits include the iPhone and iPad, will now lead design at OpenAI as the company pushes deeper into hardware. The move highlights a trend of VC-backed companies buying one another amid a shifting tech landscape and a hunger for talent. However, it is worth noting that this is the largest private-to-private acquisition ever at $6.4 billion. Indeed, over 40% (7) of all-time $1B+ private-to-private acquisitions have happened in just the last year. OpenAI, Databricks, and Stripe have each spent over 15% of their total funding to date on acquisitions in the last 2 years. Don’t forget Anja too. Venture capital investment in humanoid robots are estimated to double this year to over $2 billion per CB Insights data. Then consider that there are 660 million people in Asia (average age 27) using digital companions. That disturbing little gem came from anthropologist, Dr Lollie Mancey, in a recent RTE interview and….. I just don’t know. I’m not alone.
The fascinating story of Irish recycling software company, AMCS, and its $2 billion wealth creation story was told by its founder, Jimmy Martin, at the Renatus/Fitzgerald Power “Real Deal” SME conference in Goffs this week. When asked about AI, he wisely declined to predict the future but did make one very interesting and more definitive point. As a hugely successful observer of ‘margin’ in industry ecosystems, Martin was quick to identify the monopolistic power of the big 3 cloud infrastructure players, Microsoft, Amazon and Google. For me, the unanswered question of who will be the winner(s) will focus on the following :
- The manufacturers of the critical semiconductor chips
- The owners of data centre infrastructure
- The providers of energy/power capacity
- Sovereign/digital alignment (China, Europe or US).
I really don’t know, particularly the geopolitical/sovereign and energy/power questions. However, I do think it interesting that in recent days companies exposed to the nuclear power industry have seen big share price moves. Not coincidentally, the US and a number of European countries have been embracing a nuclear industry revival at the same time. Plenty to ponder, not all of it comfortable. Isn’t that right, Anja?