AI’s Global Capital Race Accelerates

**SPEAKER_1** (0:02)
Bloomberg Audio Studios, Podcasts, Radio, News. Bloomberg Tech is live from the heart of Silicon Valley with Ed Ludlow in San Francisco.

**Ed Ludlow** (0:22)
This is Bloomberg Tech coming up. All eyes on Nvidia reporting earnings this week, and markets are mixed. As Bloomberg reports, Nvidia customers have been notified about price hikes, plus SoftBank plans a record $6.3 billion retail bond sale, the biggest by any issuer in Japan, to raise funds for its investment commitments to OpenAI.
And Alibaba's raised $10.2 billion in what was Hong Kong's biggest follow-on offering to take the lead in the global AI race. Let's get right to our top story. Bloomberg reporting over the weekend that Nvidia-based AI servers are about to get more expensive. Server makers are telling data center customers that prices are going up more than 15% in some cases, according to sources who have seen those communications. The reason, soaring memory costs. Sources say the increase has hit systems containing Nvidia's Blackwell and next-generation ViroRubin platforms. Heights expected on shipments starting in early 2027 How are markets reacting? Very simply, AI infrastructure stocks are falling. Nvidia is down significantly. A seventh straight drop is the longest losing streak since September of 2022 Nvidia earnings Wednesday. Bloomberg's Carmen Reineke here with Markets. Carmen, what is driving investor sentiment towards this space right now?

**Carmen Reineke** (1:40)
Yeah, so we're seeing a lot of selling pressure here. And there's just so much caution, I think, baked into this space. We've seen so much volatility in the AI trade overall. And Nvidia is the kingmaker here, right? So a seventh losing day, that's the longest streak, as you said, since 2022, heading into earnings on Wednesday. People are worried here about peak earnings. Memory costs are high. Nvidia's looking like it's not absorbing those, it's passing them on to customers.
There are big questions sort of outstanding about all of the financing that it's put into the market lately. And investors really just want to hear more about what the future is of the artificial intelligence trade there. I think more and more concerns coming online that this could be the peak.

**Ed Ludlow** (2:24)
Our markets team leads its rap with this idea that price is going up for Nvidia-based servers, is something in the market. Wednesday is the mega event, right? The macro level event that is one company's quarterly earnings. How are markets kind of preparing for that?

**Carmen Reineke** (2:40)
I mean, it's such a huge deal, right? So Nvidia is still the largest stock in the S&P 500, in the NASDAQ it's the largest stock in the market. It commands a major amount of weight, and how it moves can move the entire market. The thing that's really interesting here is that earnings have not been a positive catalyst for Nvidia stock for quite some time. So you think it's the last four out of five earnings reports, the shares have actually fallen after results, even though they've been strong. What investors are really looking at is the sentiment around this reports. You know, what CEO Jensen Wong says about the market going forward, if they're seeing demand, financing questions are going to come in. I'm sure there will be questions around these price hikes as well. So a lot of that sentiment, it's really, what is the next wave of the artificial intelligence trend? That's what's going to be most important for investors coming up. And, you know, it's going to affect other parts of the AI infrastructure trade as well. You know, we're seeing the memory stocks falling today as well. Those are the biggest laggards on the S&P and the NASDAQ. That's really important.

**Ed Ludlow** (3:40)
Loomis, Carmen, Reineke, with a preview of everything to come this week, that is Nvidia. Thank you very much. Let's get more on the report. The Nvidia systems could see a price hike of 15%.
Simon Leopold, Managing Director of Data Infrastructure at Raymond James joins us now. This reporting is based on the OEMs, the server makers going to customers and saying, this is what we see happening in the new year. 15% hike in some cases on Blackwell and Rubin systems. Just to start your reaction to that.

**Simon Leopold** (4:10)
Yeah, I'm not surprised by it. It's been very clear that the cost of memory has been increasing.
We have observed and heard the large hyperscale operators forecasting higher capital spending budgets, and the biggest driver for that has been the cost of memory. So there really should be very little surprise. As a matter of fact, given the percentage increase in the cost of memory and the amount of memory required in these AI platforms, 15% really doesn't seem all that demanding when we consider the amount of spending.