It seems that 2025 was the year of the AI chip. Nvidia was a case in point. At one stage the company was worth more than $5-trillion, a world record.
Granted, it has come down substantially since then, but it is still worth an unfathomable amount in rand. Nvidia has declined in value because of moves other companies are making to beat its Blackwell chips. There’s a large amount of noise — almost hero worship — around Google’s own tensor-processing units (TPUs), used to train its new AI model and onboard major clients, including Meta Platforms.
These Gemini 3 chips are cheaper than Nvidia’s and have been received with rave reviews. Competition is so fierce in the AI space that several deals worth enormous amounts of money are in the works. Investors are buying and selling shares left, right and centre.
Read Full Article on Business Day
[paywall]
For example, Peter Thiel’s hedge fund Thiel Macro sold its entire Nvidia stake for about $100m at the end of September. Some of the recent major moves between AI companies include: Anthropic seems to have started causing waves when the deal-making began. Before that it was just the company that made Claude – an LLM based on constitutional AI, which involves training AI to follow human-defined principles.
A lot is also happening in the power space, with companies building huge nuclear plants. All of this seems to have the makings of a bubble and investors are concerned about the complicated financing methods funding it all. Others are shrugging off worries, pointing to sound balance sheets.
AI companies have manoeuvred themselves to try beat others in the race. Now we wait to see how that race will turn out, which feels almost like an interlude between two parts of a theatrical performance. We don’t know quite what the second act in 2026 will bring, though there are many predictions.
Gartner believes we’ll see growing adoption of integrated computing platforms combining the various chips Google and Nvidia are racing to develop, to the point where they are pushing against the laws of physics. If Gartner is right companies will be able to run high-fidelity simulations in finance, logistics, supply chain optimisation or climate modelling. Imagine, as a CFO, being able to predict revenue for the financial year with granular accuracy and influence strategy based on this detailed information.
More importantly, we’d be able to predict droughts and floods, which is crucial for a world in which millions go to bed hungry because climate change means crop planting and harvesting is hit and miss. Intelligent machines and robots may become a reality. Not quite a Jetsons scenario where “Rosie the Robot” brings you your cocktail, but autonomous devices that sense, reason and act in warehouses, logistics, security and manufacturing.
Other advances include the convergence of AI with biotech, bringing breakthroughs in drug design, disease detection and tailored therapies. This holds enormous potential for AI to be used for good, though its involvement in health would have spillover effects we need to solve for. If people live longer how are they going to financially support themselves?
[/paywall]
