BIG TECH TALKS—Justifying the biggest capex spend in history
I have put together quotes from the last earnings call on what Big Tech are seeing and why they are investing. MSFT META AMZN GOOG. Along with deciding what to do about the software route, this is one of the biggies in the investing world
CONTEXT
Below are selective quotes from the four big AI spenders' recent results. MSFT is perhaps the most conservative and has to manage the OpenAI relationship that is a double-edged sword for them. Without a huge in-house AI pedigree, they have the challenge of staying relevant to their large commercial customer base. Working to embed an interface between clients and LLMs is how they plan to defend their turf. Meta is perhaps the most zealous and effusive for AI, being founder-led. Meta is also perhaps the company that has proven the AI use case in an operational sense, with spectacular profit growth in its ads business. As per below, they are enthusiastic to continue the progress on the core ads business and are also spending on personal superintelligence, which is more speculative. GOOG are historically perhaps the most guarded in comments. However, they have the full AI stack and can invest across the chain. GOOG's ambitions are across the board, consumer and corporate. AMZN I find the most open and forthcoming with detail. Again, a big leading cloud business and a big retail business with optionality for AI uses. The challenge is to keep its incumbent AWS business as the leader, with the focus moving from cloud to AI uses.
SUMMARY
All these players have common themes. Demand is immense, supply is constrained, monetising is occurring and encouraging them to invest further. They also believe that the future demand is much more immense, and it is only a matter of time for all companies to move to where the leaders are now (Jassey (AMZN) gives the best explanation below—the barbell analogy). It is a once-in-a-generation opportunity, and they all want their part in it. Management would see the return on vintage capex, which they imply, and we can calculate, is attractive. ROIC is falling for all of them, but if we compare current returns with past capex, the ROIC is much better. That is, ROIC is falling due to the large increase in the denominator. Will the numerator catch up? Can the returns hold as the capex balloons in size? Is the big question. The answer boils down to incremental returns on capex, as it almost always does.
I tried to put these numbers in context. It would require a 3-5% eps increase for the entire S&P500 for this capex to earn 15-20% returns (some broad assumptions here) and all be paid to the providers. That’s big.
MSFT
As agents proliferate, every customer will need new ways to deploy, manage and protect them. We believe this creates a major new category and significant growth opportunity for us.
we continue to see strong demand across workloads, customer segments and geographic regions, and demand continues to exceed available supply.
And the way to think about that is the majority of the capital that we're spending today, and a lot of the GPUs that we're buying are already contracted for most of their useful life. And so a way to think about that is much of that risk that I think you're pointing to isn't there, because they're already sold for the entirety of their useful life. And so part of it exists because you have this shorter-dated RPO because of some of the M365 stuff. If you look at the Azure only, RPO is a little bit more extended. A lot of that is CPU basis. It's not just GPU. And on the GPU contracts that we've talked about, including for some of our largest customers, those are sold for the entire useful life of the GPU…………is that as you go through the useful life, actually, you get more and more and more efficient at delivery. So where you've sold the entirety of its life, the margins actually improved with time.
And when you think about that portion (non OpenAI Azure backlog) alone growing 28%, it's really impressive work on the breadth as well as the adoption curve that we're seeing, which is I think what I get asked most frequently, it's grown by customer segment, by industry and by geo. And so it's very consistent
That's, I think, the most magical thing, which is you deploy these things. And suddenly, the agents are helping you coordinate, bring more leverage to your enterprise.
Then on top of it, of course, there is the transformation, which is what businesses are doing. How should we think about customer service. How should we think about marketing. How should we think about finance. How should we think about that and build our own agents.
META
Our vision is building personal super intelligence. We're starting to see the promise of AI that understands our personal context, including our history, our interests, our content and our relationships. A lot of what makes agents valuable is the unique context that they can see. And we believe that Meta will be able to provide a uniquely personal experience.
But soon, we'll be able to understand people's unique personal goals, and tailor feeds to show each person content that helps them improve their lives in the ways that they want
Our feeds will become more interactive overall. Today, our apps feel like algorithms that recommend content. Soon, you'll open our apps, and you'll have an AI that understands you and also happens to be able to show you great content or even generate great personalised content for you.
We're architecting our systems so that we can be flexible in the systems that we use, and we expect the cost per gigawatt to decrease significantly over time through optimising both our technology and supply chain.
while continuing to make our systems more responsive to people's real-time interests. We're also focused on incorporating LLMs to understand content more deeply across our platform, which will enable more personalised recommendations.
We're seeing in our early testing that personalised responses drive higher levels of engagement, and we expect to significantly advance the personalisation of Meta AI this year. This dovetails with our investments in content understanding
Since the beginning of 2025, we've seen a 30% increase in output per engineer, with the majority of that growth coming from the adoption of agentic coding, which saw a big jump in Q4. We're seeing even stronger gains with power users of AI coding tools, whose output has increased 80% year-over-year. We expect this growth to accelerate through the next half.
We think that there are going to be opportunities, both in terms of subscriptions and advertising and all of the different things that you see on that.
so that way we can just offer more integrated solutions for the many, many millions of businesses that use and rely on our platforms, which is going to be really powerful, both for accelerating their results using the existing products that we have, and I think adding new lines as well
So we expect over the course of 2026 to have significantly more capacity this year as we add cloud. But we'll likely still be constrained through much of 2026 until additional capacity from our own facilities comes online later in the year
I think the important thing is we're not just launching one thing, and we're building a lot of things. I think they're -- like AI is going to enable a lot of new experiences. I outlined thematically a bunch of these in the upfront comments around personal AI around LLMs combining with the recommendation systems………There are all these different things as well as several things that we think are new that we're going to try that are not just extensions of the current things that we're doing.
But it I just think the fact that agents are really starting to work now is quite profound. And I think it is going to allow -- we're already starting to see the people who adopt them are just being significantly more productive. And there's a big delta between the people who do it and do it well and the people who don't. And I think that's going to just be a very profound dynamic for, I think, across the whole sector and probably the whole economy going forward in terms of the productivity and efficiency with which we can run these companies, which I think -- my hope is that we can use that to just get a lot more done than we were able to before
This is the first time we have found a recommendation model architecture that can scale with similar efficiency as LLMs. And we're hoping that this will unlock the ability for us to significantly scale up the size of our ranking models while preserving an attractive ROI
As we scale, we are getting dramatically more efficient. We were able to lower the Gemini serving unit cost by 78% over 2025 through model optimizations, efficiency and utilization improvements.
Our first-party models like Gemini now process over 10 billion tokens per minute via direct API used by our customers, up from 7 billion last quarter
Our 10-year track record in building our own accelerators with expertise in chips, systems, networking and software translates to leading power and performance efficiency for large-scale inference and training.
We're investing in AI compute capacity to support Frontier model development by Google DeepMind, ongoing efforts to improve the user experience and drive higher advertiser ROI in Google Services, significant cloud customer demand as well as strategic investments in Other Bets (Waymo)
I expect the demand we are seeing across the board across our services, what we need to invest for future work for Google DeepMind as well as for cloud, I think, is exceptionally strong. And so I do expect to go through the year in a supply-constrained way
It was exciting to see the fact that we're already monetizing and you saw it in the results that we just issued this quarter, the investments that we've made in AI.
It's already delivering results across the business. I know it in cloud, it's very obvious external, but you've heard the comments on the success we're seeing in search,
But I think specifically at this moment, maybe the top question is definitely around compute capacity, all the constraints, be it power, land, supply chain constraints, how do you ramp up to meet this extraordinary demand for this moment, get our investments right for the long term and do it all in a way that we are driving efficiencies and doing it in a world-class way.
approximately 60% of our investment in 2025, and it's going to be fairly similar in 2026, went towards machines, so the servers. And then 40% is what you referred to as long-duration assets (land and buildings)
We see AI Overviews and AI Mode continue to drive greater search usage and growth in overall queries, including important ones in commercial queries. Gemini-based improvements in search ads help us better match queries and craft creatives for advertisers. I talked about the understanding of intent and how this has significantly expanded our ability to deliver ads on longer and more complex searches that were, frankly, previously difficult to monetise. AI Max, for example, is already used by hundreds of thousands of advertisers and continues to unlock billions of net new queries in that sense. We see strength with SMB advertisers expanding their budgets and adopting automation tools, leading to better ROI. On the creative side, we're using Gemini to generate millions of creative assets via text customisation in AI Max and PMax and so on. So we're very pleased with what we're seeing here.
AMZN
We're seeing strong growth and with the incremental opportunities available to us in areas like AI, chips, low earth orbit satellites, quick commerce and serving more consumers' everyday essentials needs, we have a chance to build an even more meaningful business in Amazon in the coming years with strong return on invested capital, and we're investing to do so.
We're continuing to see strong growth in core non-AI workloads as enterprises return to focusing on moving infrastructure from on-premises to the cloud
We expect to invest about $200 billion in capital expenditures across Amazon, but predominantly in AWS because we have very high demand, customers really want AWS for core and AI workloads, and we're monetising capacity as fast as we can install it. We have deep experience in understanding demand signals in the AWS business and then turning that capacity into a strong return on invested capital. We're confident this will be the case here as well.
We are putting into service with customers all the capacity that we're getting, and it's immediately useful. And we're also seeing a long arc of additional revenue that we see from other customers and backlog and commitments that people are anxious to make with us, especially for AI services. We see a strong return on invested capital. We see strong demand for these services, and we continue to like the investments in this area.
I would add to that. If you look at the capital we're spending and intend to spend this year, it's predominantly in AWS. And some of it is for our core workloads, which are non-AI workloads because they're growing at a faster rate than we anticipated. But most of it is in AI. And we just have a lot of growth and a lot of demand…….And what we're continuing to see is that as fast as we install this capacity, this AI capacity, we are monetising it. And so it's just a very unusual opportunity.
And I think the other thing is that if you really want to use AI in an expansive way, you need your data in the cloud, and you need your applications in the cloud. Those are all big tailwinds pushing people towards the cloud. So we're going to invest aggressively here, and we're going to invest to be the leader in this space as we have been for the last number of years
I'm very confident we're going to have a strong return on invested capital here.
(below best describes what they see happening)
The way I would describe what we see right now in the AI space is it's really kind of a barbelled market demand where on one end, you have the AI labs who are spending gobs and gobs of compute right now, along with what I would consider a couple of runaway applications (Claude, ChatGPT). And then at the other side of the barbell, you've got a lot of enterprises that are getting value out of AI in doing productivity and cost avoidance types of workloads. These are things like customer service, business process automation or some of the fraud pieces. And then in the middle of the barbell are all the enterprise production workloads. And I would say that the enterprises are in various stages at this point of evaluating how to move those, working on moving those and then putting them into production. But I think that the middle part of the barbell very well may end up being the largest and the most durable. And I would put in the middle of that barbell, too, by the way, I would put just the altogether brand-new businesses and applications that companies build that right from the get-go run in production on top of AI.
And so I think that to me, when I look at this, and what's happening, it's kind of unbelievable if you look at the demand of what you're seeing already with AI, but the lion's share of that demand is still yet to come in the middle of that barbell. And that will come over time. It will come as you have more and more companies with AI talent as more and more people get educated with an AI background. As inference continues to get less expensive, and that's a big piece of what we're trying to do with Trainium and our hardware strategy. And as companies start to have success in moving those workloads to -- further and further success in moving those workloads to run on top of AI. So I think there's -- it's just a huge opportunity. It's still in the relatively early stages, even though it's growing at a very -- like an unprecedented clip as we've talked about.
So we're growing at a really unprecedented rate, yet I think every provider would tell you, including us, that we could actually grow faster if we had all the supply that we could take.
Disclosure I hold all these companies.