Products You May Like
AI hype and data-center capex have been major market drivers in the past year but there is some angst lately about how much is being spent and whether there is a real business case for it all. That really kicked off with DeepSeek undercutting cost assumptions and now there is a TD note doing the rounds suggesting that Microsoft is pulling back on spending.
I’ve published the text of the note below.
Our channel checks indicate that $MSFT has 1) canceled leases in the US, totaling ‘a couple of hundred MWs’ with at least two private data center operators, 2) has pulled back on the conversion of 500’s to leases, and 3) has re-allocated a considerable portion of its international spend to the US. When coupled with our prior channel checks, it points to a potential oversupply position for MSFT.
Our Channel Checks Indicate Microsoft Has Cancelled Select US Data Center Leases and Has Pulled Back on SQQ To Lease Conversion; Pulls Back on International Market Expansion.
Our recent channel checks indicate that Microsoft has terminated select leases with at least two private data center operators across multiple U.S. markets, totaling a couple of hundred MW. Our checks indicate that in some situations, Microsoft is using facility/power delays as a justification for the termination. Recall, as we highlighted in our 2022 Takeaways from PTC, this is the same tactic that Meta used to cancel multiple data center leases in the U.S. after we learned in our checks that Meta had then canceled a $48B capex program related to the metaverse (Meta subsequently cut its capex guidance by $5.4B two weeks later).
Separately, our channel checks suggest that Microsoft has also pulled back on converting negotiated and signed Statement of Qualifications (500’s) (the precursor to a data center lease) into signed leases. To this point, it is currently unclear to us if this is simply a delay in 500-to-lease conversion or if it is an outright termination of the 500 with no conversion to leasing expected. For context, based on our checks, a 500 sets forth the terms and conditions for the lease and does not constitute a lease agreement. However, the conversion rate of 500’s into a signed lease is close to 100%, with data center operators using this as the signal to start data center construction. In addition, our channel checks indicate that Microsoft is also re-allocating a considerable portion of its projected international spend to the U.S., which suggests to us a material slowdown in international leasing.
Why is this happening?
While we have yet to get the level of color via our channel checks that we would like into why this is occurring, our initial reaction is that this is tied to Microsoft potentially being in an oversupply position. As we highlighted in our recent takeaways from PTC, we learned via our channel checks that Microsoft:
1. Walked away from multiple 100MW deals in multiple markets that were in early/mid-stages of negotiations.
2. Let +1GW of LOIs on larger footprint sites expire.
3. Walked away from at least five land parcels that it had under contract in multiple Tier I markets.
At the time, we also highlighted that the magnitude of both potential data center capacity it walked away from and the decision to pull back on land acquisition (which supports core long-term capacity growth) in our view indicates the loss of a major demand signal that Microsoft was originally responding to and that we believed the shift in their appetite for capacity is tied to OpenAI, which recent press reports appear to confirm.
To that point, consider this: Microsoft was the most active lessee of capacity in 2023 and 1H24, at which time it was procuring capacity relative to a capacity forecast that contemplated incremental OpenAI workloads. However, as we believe is indicated by its decision to pause construction on a data center in Wisconsin—which our prior channel checks indicated was to support OpenAI—there is capacity that it has likely procured, particularly in areas where capacity is not fungible to cloud, where the company may have excess data center capacity relative to its new forecast.
This is our interpretation of the current situation, however, our view is subject to change as we conduct incremental channel checks.”
There is has long been some angst about AI spend slowing down and signs the ‘bubble’ is deflating. However contrast this with what CEO Satya Nadalla said in a podcast this week.
So I think there are two places where I can say with some confidence. One is the hyperscalers
that do well, because the fundamental thing is if you sort of go back
to even how Sam and others describe it, if intelligence is log of
compute, whoever can do lots of compute is a big winner.
So in fact it’s manna from heaven to have these AI workloads
because guess what? They’re more hungry for more compute, not just for
training, but we now know, for test time. When you think of an AI agent,
it turns out the AI agent is going to exponentially increase compute
usage because you’re not even bound by just one human invoking a
program. It’s one human invoking programs that invoke lots more
programs. That’s going to create massive, massive demand and scale for
compute infrastructure. So our hyperscale business, Azure business, and other hyperscalers, I think that’s a big thing.
There are also some indications around a breakup of the Microsoft and OpenAI partnership.
It’s tough to tell what’s true and what’s true and what’s regular churn here but there is enough noise around this to add to the angst early this week.