00:00 Speaker A
A busy day closing deals in the AI space. Open AI added to list of circular transactions. The Chat GPT maker has now taken ownership of Thrive Holdings, a company founded by one of Open AI’s investors. Open AI has also signed an agreement to become one of Accenture’s key AI partners. Nvidia is also doubling down on AI deals, announcing Monday that it invested $2 billion in Synopsys.
00:27 Speaker A
So, with AI training well under way, we look at how our deals and investments could turn out with Kunjan Shabanini, Senior Semiconductor Analyst at Bloomberg Intelligence. And Kunjan, we talk a lot about whether we’re in an AI bubble, valuation, part of that story, but we also talk about circular investing, which we talked about earlier. So how do you see it? Or does that not matter at all given its vitality for now, the continuation of AI?
01:07 Kunjan Shahani
Now, so far, if we look at the fundamentals in 2026, hyperscalar capex is projected to exceed $5 trillion. And remember, these are customers who have excess balance sheet cash. So you don’t have to worry about whether they can use this money or not. Fundamentals suggest that there will be no slowdown in demand and spending for AI chips, at least in 2026, especially for GPUs and AI ASIC chips, and not until the first half of 2027.
01:42 Kunjan Shahani
Here are some key items I would like to highlight as we head into 2026. We again think Nvidia is likely to maintain a dominant wallet share of this spending, but the second half of 2026 will be a key inflection point for other players. The first is AMD. This is when the company’s first server rack-level solutions begin to be deployed in data centers. Accordingly, we expect to see large-scale AI-based accelerator deployments from both Google and AWS in the second half of 2026. This is the first time we have introduced it to an external customer. Therefore, many eyes will be focused on these lamps.
02:11 Speaker A
And it’s interesting that you talk about the different paths these different companies could take. Because if you look at the Mac 7 players right now, there’s obvious variation. They no longer necessarily move in lockstep. So what does this say about where we are in the current AI cycle?
02:29 Kunjan Shahani
Well, there are just more vectors to drive spending now. Mac 7 and the largest hyperscale consumption cloud service providers have significant and verified demand and spending capacity. And then there are also new hyperscalars that are emerging, like Open AI, right? So a lot of the concerns around this revolving financing and recycling revenue and trading are coming from new parts of the hyperscalar that are also growing, so they’re going to have a significant impact.
03:02 Kunjan Shahani
However, it will take at least the next 12 to 18 months to see any real impact. So at this point, it’s more of an outwardly directed concern.
