Microsoft expects to
spend more than US$50 billion on capital expenditure in the December
quarter of 2026, up from US$45.2 billion in the September quarter.
Alphabet also plans to increase investment significantly in 2027. For
Nvidia, these plans indicate continued appetite for the infrastructure
used to deliver artificial intelligence (AI) services. Its next-quarter
sales, however, also depend on how much equipment it can deliver.
Nvidia reports on 17
November for its fiscal third quarter, which ended on 25 October 2026.
The cloud providers’ September results offer evidence about an overlapping
period. Their spending plans look further ahead, helping traders assess
whether investment is likely to continue beyond the quarter Nvidia is
about to report.
What the Spending
Plans Tell Us
Microsoft’s
December-quarter forecast provides a near-term indication of its
investment intentions. Further spending is expected in the first half of
calendar 2027, including some third-party computing capacity whose
delivery was delayed. Its plans cover both central processing units
(CPUs) and graphics processing units (GPUs), alongside higher memory and
storage costs.
Alphabet retained its
capital expenditure forecast of US$195–205 billion for 2026 and said spending
would increase significantly in 2027. Amazon also maintained its forecast
of approximately US$220 billion in cash capital expenditure for 2026.
These company-wide budgets use different accounting measures and cover
much more than Nvidia equipment.
The timing limits what
those figures tell us about the next earnings announcement. Microsoft’s
December-quarter plan is more relevant to the immediate outlook than
Alphabet’s investment intentions for the whole of 2027. The longer-term plans instead
offer evidence of how far cloud providers expect to continue expanding.
How Much Spending
Reaches Nvidia?
The September-quarter
breakdowns help explain where the money goes. Microsoft allocated roughly
three-quarters of its capital expenditure to shorter-lived assets,
primarily CPUs and GPUs. Its US$45.2 billion total included US$7.4
billion of finance leases, so the figure should not be read as cash paid
for new chips during the quarter.
Separately, businesses
can rent computing capacity from providers such as CoreWeave, which
operates Nvidia infrastructure. In that arrangement, the business pays to
use computing resources while the provider acquires the equipment.
Spending on cloud services can therefore support Nvidia purchases
indirectly.
Alphabet spent US$51.4
billion, mostly on technical infrastructure. Approximately 65% of that
infrastructure investment went towards servers and networking equipment, with
the remaining 35% allocated to data centres. The equipment portion is
more relevant to Nvidia, which supplies computing systems and networking
technology alongside processors.
Cloud providers also
have alternatives. Alphabet uses its own tensor processing units (TPUs)
alongside Nvidia equipment, while Amazon has reported strong demand for
its Trainium chips. Their budgets therefore leave open how purchases are
divided between suppliers. Nvidia’s wider customer base also includes
specialist cloud providers, enterprises, and government-backed AI
projects, so these three companies provide only part of the demand
picture.
Supply Limits
Near-Term Deliveries
On its August earnings
call, Nvidia described demand exceeding available supply. In October, it
announced further capacity expansion plans with manufacturing partners,
covering advanced packaging and other production stages.
Two constraints
highlighted by Parker Tian are high-bandwidth memory (HBM), which feeds
data rapidly to AI processors, and advanced packaging, which connects
processors and memory within a chip assembly.
“I read Big Tech’s
capex as a signal of how long this cycle can last, not a forecast of
Nvidia’s next quarter, which is still set by HBM memory and advanced
packaging capacity,” says Parker Tian, Head of EBC AI Lab at EBC
Financial Group.
Nvidia has described
pressures across its wider supply chain too. Expanding capacity takes
time, so larger cloud budgets do not necessarily translate into an
immediate increase in equipment available for sale.
Higher Costs Also
Affect the Outlook
Supply pressures
affect profitability as well as delivery volumes. Nvidia said memory
prices had risen more than it expected. Selling more systems can therefore
increase revenue even as higher component costs reduce the share retained
as gross profit.
Nvidia had already
allowed for some pressure. It guided third-quarter gross margin to 74%,
plus or minus half a percentage point, compared with 75% in the preceding
quarter. Management also projected a further decline to 71–72% in the
fourth quarter. A lower margin should therefore be compared with those
forecasts before being interpreted as an additional setback.
What to Look for on 17
November
Nvidia’s revenue
guidance, issued on 26 August, was US$108 billion, plus or minus 2%, with
no assumed data-centre compute revenue from China. Investing.com displayed
an analyst forecast of US$108.8 billion when checked on 13 November.
Revenue at management’s midpoint would meet its guidance while falling
slightly below that published forecast.
The following
quarter’s guidance will show how much revenue Nvidia expects to generate,
while its margin outlook can indicate how component costs are affecting
profitability. If revenue guidance disappoints, management’s explanation
will be needed to distinguish production delays from weaker orders.
“The warning sign
could be management describing orders, rather than supply, as the
constraint, which would mean the AI build-out is shifting from a supply problem
to a demand question,” says Tian.
This article was written by IL Contributors at investinglive.com.