The AI spending spree has a new headline number. A Morgan Stanley report projects global AI-related capital expenditures will hit approximately $1.58 trillion by 2027, a figure often rounded to $1.6 trillion.
The more striking part is where that money is expected to land. Nvidia could collect approximately $690 billion of it, rounded to $700 billion in some assessments. That would make one chipmaker the biggest beneficiary of one of the largest corporate building cycles on record.
Following the money
Morgan Stanley’s math works like a funnel. Roughly 65% of total AI capex, or about $1.03 trillion, is expected to flow to companies in the S&P 500.
Nvidia is then expected to take about two-thirds of that S&P 500 share.
The forecast also got bigger recently. Morgan Stanley raised its 2027 total by about 14% after accounting for lease obligations that had not previously been included in the count.
Leases are a quieter form of spending. A company that rents data center capacity instead of building it still commits real money to AI, it just shows up in a different line of the financial statements. Once Morgan Stanley started counting those commitments, the total climbed.
By category, semiconductors and servers are projected to account for about 42% of 2027 AI capex, or approximately $672 billion.
Nvidia’s own numbers point the same way
The bank’s projection lines up with what Nvidia itself has been telling investors. On the company’s fiscal Q2 2027 earnings call, CFO Colette Kress indicated expectations that hyperscaler capex would approach $800 billion in 2026 and reach around $1.3 trillion in 2027.
She also pointed to a cloud industry backlog exceeding $2 trillion.
The hyperscalers doing the heavy lifting are familiar names: Microsoft, Alphabet, Amazon, Meta and Oracle. These companies build and run the giant data centers that train and serve AI models, and Nvidia’s GPUs sit at the center of most of that buildout.
Nvidia’s chips and associated platforms are considered critical to this infrastructure spending. Upcoming technology like its Blackwell platform is a key part of how the company expects to keep serving that demand.
Not everyone agrees on the near-term pace. Estimates for 2026 hyperscaler capex have ranged from about $650 billion to more than $800 billion, with forecasters adjusting their figures throughout the year.
Why the S&P 500 cares
Morgan Stanley expects AI-related capex to contribute about 5% of S&P 500 revenue in 2027.
AI capex is also expected to drive about 25% of the index’s sales growth that year.
Earnings tell a similar story. AI “enablers” and “adopters” are expected to drive 78% of S&P 500 earnings growth in 2027.
What this means
For investors, the clearest takeaway is concentration. If Nvidia does capture approximately $690 billion, a single company would be soaking up a substantial portion of the AI capex wave.
The lease adjustment is also worth watching. A 14% upward revision driven by commitments that had previously gone uncounted suggests the true scale of AI spending may be harder to track than headline capex figures imply.
The more than $2 trillion cloud backlog Kress cited offers some reassurance, since it reflects demand customers have already committed to.
Key things to track from here: whether hyperscaler capex guidance continues to move higher, how quickly Blackwell and related platforms ship, and whether Morgan Stanley and other forecasters revise their 2027 totals again.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.