Amazon shares moved lower after the company forecast a sharp increase in capital spending, much of it tied to artificial intelligence infrastructure, reviving debate over whether Big Tech’s AI investments are rising faster than near-term returns. The company’s capex outlook came in well above expectations, adding to broader market concerns around the scale and timing of AI payoffs.
BNN Bloomberg spoke with Lloyd Walmsley, managing director and senior equity research analyst at Mizuho Americas, about signs of improving cloud momentum, how Amazon compares with rivals in AI infrastructure, and whether accelerating demand can justify elevated spending levels.
Key Takeaways
- Amazon’s higher-than-expected capital spending has renewed investor scrutiny over how quickly AI investments can generate meaningful returns.
- Cloud growth acceleration and faster backlog expansion suggest demand for AI infrastructure is improving after a slower start.
- Despite recent progress, Amazon still trails rivals in custom AI chip development and overall AI cloud maturity.
- Competitive advantages in AI increasingly depend on a mix of infrastructure, proprietary chips and data scale.
- Advertising remains one of the clearest near-term beneficiaries of AI, helping offset pressure from heavy infrastructure spending.
Read the full transcript below:
ANDREW: Amazon plans to spend US$200 billion. Now, not all of that is linked to AI, but a huge amount will be on data centres, chips and other equipment, and that capex forecast is US$50 billion higher than analysts had expected. We have seen some pressure on Amazon stock today. Let’s get more from Lloyd Walmsley, managing director and senior equity research analyst at Mizuho Americas. Lloyd, thanks very much indeed for joining us. What was your reaction? Is there a risk Amazon is overdoing it here?
LLOYD: Thank you for having me. Look, I think certainly there’s a risk everybody’s overdoing it here, but coming out of this quarter, we got a lot of what we’ve been wanting to see out of Amazon for much of the last year. We saw a market acceleration in growth at AWS, a four percentage point acceleration, which was nicely better than we had been hoping for, and backlog growth accelerated by 16 percentage points.
So they are clearly in a better position for cloud related to AI than I think people were giving them credit for just a few months ago. They are investing a lot — these numbers are eye-popping across the board — but we are seeing good signs that there is early ROI. It’s not where it’s historically been, but over time we think it can get there, and it’s a worthy investment for them to be making.
ANDREW: So you say that AWS is catching up in AI cloud. Have they, to some extent, fallen behind their big rivals?
LLOYD: Yeah, I think they had fallen behind. They were slow to get going, whether that’s building large language models themselves or building out the infrastructure. So I think they were behind for a bit. They may well still be a little bit behind some of their peers.
Google has been building its own custom AI ASIC chip for about 10 years now, versus AWS just the last few years on its Trainium chip. So they are behind, but they’re making quick progress, which is what it looks like to us.
ANDREW: One number you highlight is AWS backlog growth of US$44 billion in the fourth quarter, which really lagged Google’s backlog growth north of US$80 billion.
LLOYD: It did. The print we saw out of Google was extraordinary. The acceleration in cloud and the margin improvement were stunning. Google isn’t disclosing exactly what’s going on, but we think they’re benefiting from the move to outright sell their TPU chip.
We don’t know exactly how revenue recognition is working, but clearly what they have in that chip is something special in terms of cost performance and competitive positioning, and they’re taking advantage of that.
AWS has an incredible semiconductor team and has had enormous success historically with its Graviton chip. But on training chips, AWS is just rolling out its third generation, while Google is on something like its 10th generation of TPU chip. So Google is clearly benefiting from that lead.
ANDREW: They also have many other revenue streams. They’re making a lot of money from advertising through the retail platform as well.
LLOYD: That’s right. Advertising is one of the biggest beneficiaries of AI, and we see that across the space. Meta is an incredible example, with massive acceleration in both user engagement and ad performance. Google benefits from that too.
I don’t think Google benefits quite as much as Meta, because search has historically been such a strong ad business already, so there’s less room for improvement. But YouTube benefits tremendously, and that helps the core business.
Google is unique in that it’s benefiting on the application side with advertising, the infrastructure side with TPUs, and it has so much proprietary data to inform its models that it’s in a uniquely strong AI position.
ANDREW: Lloyd, thank you very much indeed for joining us. That’s Lloyd Walmsley, managing director and senior equity research analyst at Mizuho Americas, with an outperform rating and a US$315 price target.
---
This BNN Bloomberg summary and transcript of the Feb. 6, 2026 interview with Lloyd Walmsley are published with the assistance of AI. Original research, interview questions and added context was created by BNN Bloomberg journalists. An editor also reviewed this material before it was published to ensure its accuracy and adherence with BNN Bloomberg editorial policies and standards.
