CNBC reported on 28 October 2026 that around 75% of Microsoft’s capital expenditures are now tied to short-lived assets such as Intel chips and Nvidia graphics processing units, up from about 66% in the past three quarters and half in the year-ago quarter.
CNBC put Microsoft’s capital expenditures and finance leases at $45.2 billion for the quarter to 30 September 2026. By our arithmetic, 75% of that is about $34 billion of short-lived chips and servers in a single quarter.
Business Pill · Capex and Depreciation
An evergreen Business Engineer explainer on capital expenditure and depreciation. It explains the general idea only and does not discuss Microsoft’s or Alphabet’s results.
The key insight: By our arithmetic from CNBC’s report, about $34 billion of Microsoft’s $45.2 billion of quarterly capital expenditures and finance leases went to short-lived assets such as chips and GPUs, around 75%, against about half a year earlier.
CNBC said the $45.2 billion was up about 30% from a year earlier, against the $50.67 billion analysts surveyed by StreetAccount had projected.
Microsoft’s quarterly report filed on 28 October 2026 shows additions to property and equipment, the cash line, of $39.4 billion, against $19.4 billion a year earlier. The same report shows right-of-use assets obtained under finance leases of $5.2 billion, against $9.1 billion a year earlier.
CNBC wrote that Microsoft flagged a higher growth rate for cash paid for property and equipment than for overall capital expenditures. It quoted finance chief Amy Hood on the conference call: “Some third-party capacity deliveries we expected in Q1 were delayed”.
Share of quarterly capital spending going to short-lived servers and chips. Microsoft shares as reported by CNBC (capital expenditures including finance leases); Alphabet shares as stated by its CFO on the Q3 2026 call (technical infrastructure investment). The two bases differ; the $34 billion estimate is our arithmetic.
How Long the Gear Lasts
Microsoft’s fiscal 2026 annual report gives an estimated useful life of two to six years for servers and network equipment, against five to 15 years for buildings and improvements.
The quarterly report puts depreciation expense at $11.1 billion, against $7.1 billion a year earlier. By our arithmetic, the roughly $34 billion of short-lived assets bought in the quarter is about three times that quarter’s depreciation expense.
Alphabet’s Mix
Alphabet’s finance chief Anat Ashkenazi said on the company’s Q3 2026 earnings call that capex was $51.4 billion, “with the vast majority of the spend in technical infrastructure to support our investments in AI”.
She said approximately 65% of the technical infrastructure investment was related to servers and networking equipment, “which generally have useful lives of six years”, and the remaining 35% was for data centers, “which have useful lives of seven to forty years”.
Alphabet’s release puts purchases of property and equipment at $51.4 billion, against $24.0 billion a year earlier. By our arithmetic, 65% of the full $51.4 billion would be about $33 billion; because Alphabet applies the share to technical infrastructure rather than all capex, that is an upper bound.
The two measures differ: Microsoft’s share, as CNBC reported it, applies to capital expenditures including finance leases, and Alphabet’s to its technical infrastructure investment.
Update, 28 Oct: the call confirms the $45.2 billion and the three-quarters share. A machine transcript of Microsoft’s earnings webcast (Microsoft’s official transcript was not yet available) has Amy Hood saying: “Capital expenditures were $45.2 billion, as some third party capacity deliveries we expected in Q1 were delayed. Roughly three quarters of our CapEx was for short-lived assets, primarily CPUs and GPUs, including higher memory and storage costs.” That makes the share Microsoft’s own statement on the call, not only CNBC’s report.
The same transcript has Hood saying finance leases were $7.4 billion and cash paid for PP&E was $39.4 billion. Microsoft’s 10-Q shows $5.2 billion of right-of-use assets obtained in exchange for finance lease obligations in the quarter. We could not reconcile the $7.4 billion with that figure or with the $5.8 billion gap between $45.2 billion and $39.4 billion (our arithmetic), and the transcript is machine-made, so we treat the $7.4 billion as unconfirmed until Microsoft posts its official transcript.
As we read the figures, more of the spending now goes to shorter-lived equipment: Microsoft’s annual report gives servers and network equipment a useful life of two to six years, against five to 15 years for buildings and improvements.
By our arithmetic, the short-lived share bought in the quarter is about three times that quarter’s $11.1 billion of depreciation expense, according to the quarterly report.
Alphabet’s finance chief put servers and networking at approximately 65% of technical infrastructure investment, with data centers making up the remaining 35%; the two companies measure the share on different bases.
“with the vast majority of the spend in technical infrastructure to support our investments in AI”
Three Implications
FOR CHIP SUPPLIERS By our arithmetic, about $34 billion of Microsoft’s quarterly capital spending went to short-lived assets such as chips and GPUs, according to CNBC’s figures.
FOR THE INCOME STATEMENTMicrosoft’s quarterly report puts depreciation expense at $11.1 billion, against $7.1 billion a year earlier.
FOR WHAT IS NOT DISCLOSEDMicrosoft’s filing does not split capital expenditures by asset life; the 75% share comes from CNBC’s report.
The framework argues: “The thesis: the capex race is no longer a corporate spending cycle.”
Spending on chips and servers that last a few years has to be repeated as the equipment ages. CNBC’s figures put that share of Microsoft’s capital spending at around 75%. The framework was published in May 2026 and does not discuss these figures.
What Is Not Established
Microsoft’s quarterly report does not break capital expenditures down by asset life. The 75% share and the $45.2 billion figure come from CNBC’s report of Microsoft’s results and call; the dollar estimates and the comparison with depreciation are our arithmetic.
Neither company names how much of the short-lived spending went to any single chip supplier.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
CNBC reported that around 75% of Microsoft’s capital expenditures now go to short-lived assets such as chips and GPUs, up from half a year earlier. By our arithmetic that is about $34 billion in the quarter. Alphabet’s finance chief put its server and networking share of technical infrastructure investment at approximately 65%.
A note on sourcing. Microsoft’s capex total and asset-life share come from CNBC’s report of the 28 October 2026 results and call; cash additions, finance-lease assets, depreciation and useful lives come from Microsoft’s Form 10-Q and fiscal 2026 Form 10-K. Alphabet’s mix comes from its CFO’s remarks on the Q3 2026 earnings call and its release. Dollar estimates and ratios are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
