The wave of investment in AI infrastructure is pushing tech giants into an unprecedented heavy asset cycle. Morgan Stanley's latest research shows that hyperscalers represented by Amazon, Google, Meta, Microsoft, and Oracle are expected to have capital expenditure intensity that will fully surpass the historical peak of the internet bubble era, indicating a structural shift in the business model of the tech industry.
According to news from the Wind Trading Desk, a report released by Morgan Stanley on February 26, 2026, predicts that the capital expenditure to sales ratio (capex-to-sales) for the aforementioned five hyperscalers will reach 34%, 39%, and 37% respectively from 2026 to 2028, exceeding the peak of about 32% during the internet bubble period.
If financing leases are included in the calculation, this ratio will further rise to 38%, 44%, and 45%. Meanwhile, the total capital expenditure of these companies over the next three years will exceed $2 trillion, accounting for about 40% of the total capital expenditure of the Russell 1000 index constituents.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/17e1348e-e070-4a2e-a7a4-81e005173c51.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/17e1348e-e070-4a2e-a7a4-81e005173c51.png"/>
However, the explosive expansion of capital expenditure has not brought about a corresponding increase in revenue. Morgan Stanley pointed out that over the past six months, the market's consensus expectation for capital expenditure in 2026 to 2027 has been raised by more than $630 billion, but the revisions to revenue expectations have been much more limited, leading to a continued decline in the expected free cash flow (FCF) of hyperscalers. In contrast, semiconductor AI-enabled companies have seen their consensus revenue expectations for 2026 raised by about 60% over the past two years, far exceeding the approximately 8% increase for hyperscalers, making them the group most directly benefiting financially from this round of AI investment cycle.
<h2>Capital Intensity Breaks Historical Highs of the Internet Bubble</h2>
Morgan Stanley noted in the report that six months ago, the firm characterized the wave of AI construction as &#34;close but not yet surpassing&#34; the capital intensity peak of fiber optic construction during the internet bubble. The latest forecasts now indicate that capital intensity will &#34;far exceed&#34; the peak of about 32% during the internet bubble, with expected capex-to-sales ratios of 34%, 39%, and 37% from 2026 to 2028.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/9b1754af-d92c-4747-b987-675d9ad3529b.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/9b1754af-d92c-4747-b987-675d9ad3529b.png"/>
The report also emphasizes that measuring only traditional capital expenditure actually underestimates the scale of this investment cycle. Financing leases essentially acquire assets through debt and should be included in the assessment of total investment scale. There was almost no use of financing leases during the internet bubble era, but current hyperscalers are signing data center leases worth hundreds of billions of dollars. Morgan Stanley's software industry analysts expect that the financing lease capital expenditure of just Microsoft and Oracle will be enough to push the overall hyperscalers' capex-to-sales ratio to 38%, 44%, and 45% from 2026 to 2028
From the perspective of the impact on the Russell 1000 Index, in 2025, hyperscale cloud service providers contributed more than 150% of the capital expenditure increment of the index—indicating that the capital expenditure of the remaining constituent stocks is actually in net contraction. The capital expenditure of hyperscale cloud service providers grew by approximately 70% year-on-year, while the remaining constituent stocks of the index decreased by 6%. Morgan Stanley expects that by 2026, hyperscalers will account for about 40% of the total capital expenditure of the Russell 1000, doubling from 2024, and may further rise to 49% by 2028.
<h2>Capital Expenditure Revisions Reach Record Levels, Revenue Forecasts Lag Significantly</h2>
A notable feature of this investment cycle is the unprecedented speed and magnitude of the upward revisions to capital expenditure forecasts. Since September 2025, the market's consensus expectations for hyperscalers' capital expenditures in 2026 and 2027 have been raised by approximately 1.5 times, while Morgan Stanley's own analysts have raised their forecasts by as much as 1.8 times.
From the perspective of individual companies, Google's consensus expectation for capital expenditure in 2026 has been revised up by 117% compared to a year ago, META by 96%, Amazon by 75%, and Oracle's revision is as high as 264%. Morgan Stanley analyst Todd Castagno's team pointed out that these revisions exhibit a &#34;step-like&#34; characteristic rather than a gradual adjustment, indicating that this investment cycle is extremely difficult to predict—management continues to update data center expansion plans, and companies are competing to secure key supply chains, further complicating the forecasting process.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/81ae041e-fa92-4b92-88d4-b6d741b91102.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/81ae041e-fa92-4b92-88d4-b6d741b91102.png"/>
In stark contrast to the rapid upward revisions in capital expenditure expectations, revenue revisions have barely moved, and free cash flow (FCF) expectations have declined accordingly. Data from the report shows that in the past year, the capital expenditures of five companies have been collectively revised up by over $310 billion for 2026, while revenue revisions have only increased by about $130 billion. Morgan Stanley noted that as the fixed cost base continues to expand, the operating leverage of these companies will rise, and the sensitivity of future profits and FCF to changes in revenue expectations will also significantly increase.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6872bd73-dfbd-42fe-ac72-e95348e662b8.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/6872bd73-dfbd-42fe-ac72-e95348e662b8.png"/>
<h2>Financing Leases Significantly Amplify Actual Investment Scale</h2>
Hyperscale cloud service providers have recently significantly expanded their use of financing leases, further raising the actual capital intensity levels. As of the latest company financial reports, these five companies have total commitments for future leases exceeding $660 billion, with Oracle at approximately $248 billion, Microsoft at about $155 billion, META at around $104 billion, Amazon at about $96 billion, and Google at approximately $59 billion. Notably, Google's lease commitments have increased by about seven times since 2024, while META's increase during the same period exceeds 200%
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/765ebec6-425a-48fc-9605-abdbdf03cec0.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/765ebec6-425a-48fc-9605-abdbdf03cec0.png"/>
The impact of financing leases on the capital intensity of individual companies is particularly prominent. Taking Microsoft as an example, if only traditional capital expenditures are considered, its capex-to-sales ratio for FY26 and FY27 is approximately 29%. When financing leases are included, this ratio will jump to about 43% and 42%, respectively. The situation for Oracle is even more extreme—this company is acquiring all data center shells through leasing. Under traditional metrics, Oracle's capex-to-sales is expected to be 75% and 119% for FY26 and FY27, respectively, but when financing leases are included, it will rise to 107% and 201%, indicating that the total reinvestment scale for both fiscal years will exceed the total revenue for those years.
<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/1002d656-d0ba-4634-ba3b-19a33a3983cf.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/1002d656-d0ba-4634-ba3b-19a33a3983cf.png"/>
<h2>Semiconductor companies are the biggest winners, while returns for hyperscale cloud service providers remain to be proven</h2>
Although capital expenditures are highly concentrated among hyperscale cloud service providers, the group that has recently benefited most clearly from this investment cycle is semiconductor AI-enabled companies.
The fundamental reason for this divergence lies in the difference in revenue certainty: Hyperscale cloud service providers have made large-scale purchases of GPUs and other chip components in advance, contributing to a clear and visible revenue source for chip suppliers; however, the hyperscalers themselves need to rely on the monetization of large language models, ongoing computing power demand, and product differentiation over the next several years to gradually realize these computing power assets, which carries relatively higher uncertainty.
The performance of the capital markets also confirms this logical divergence. Since December 2023, the stock prices of North American semiconductor AI-enabled companies have outperformed those of hyperscale cloud service providers and the broader AI-enabled sector by 272% and 224%, respectively. The market is currently more willing to pay a premium for the confirmed near-term profits of semiconductor companies, while choosing to wait and see regarding the revenue realization paths for hyperscalers and the broader AI-enabled group.
Morgan Stanley analyst Brian Nowak believes that META, Google, and Amazon are accelerating user engagement and commercial monetization through AI investments, data accumulation, and scale advantages; Keith Weiss characterizes Oracle's data center expansion as a potential revenue opportunity, but emphasizes that it requires substantial financial support. The current trend of upward revisions in capex will also lead to a continuous rise in depreciation expenses, which, in the context of sales revenue not being revised upward simultaneously, will exert significant pressure on profit margins

SOXX

MSFO

Morgan Stanley's report points out that the AI arms race is pushing tech giants into a super capital cycle. The five major cloud service providers will collectively exceed the peak of the internet bubble in capital expenditure intensity from 2026 to 2028, totaling over $2 trillion in three years. Although the massive investment drives semiconductor manufacturers' revenue to soar, the giants themselves face challenges of increasing revenue without increasing profits, declining free cash flow, and significant financing lease pressures, with monetization paths still to be validated

- Morgan Stanley predicts that hyperscalers like Amazon, Google, Meta, Microsoft, and Oracle will exceed their historical capital expenditure peaks from the dot-com bubble, with capital-to-sales ratios projected to reach 34%, 39%, and 37% from 2026 to 2028.  
- The rapid rise in capital expenditures, totaling over 2 trillion USD, has not been matched by significant revenue growth, causing concerns over the companies' free cash flow forecasts.  
- Semiconductor firms have benefitted directly from AI investments, contrasting with the uncertain financial returns expected from hyperscalers as they try to monetize their significant infrastructure investments over time.

Wallstreetcn

iShares Semiconductor ETF

YieldMax MSFT Option Income Strategy ETF

Invesco Semiconductors ETF

Alibaba

Alphabet

Roundhill AMZN WeeklyPay ETF

Direxion Daily GOOGL Bull 2X Shares

SPDR S&P Semicon

Roundhill GOOGL WeeklyPay ETF

ISHRS S&P Glb It

Proshares UltraPro QQQ

VanEck Semiconductor ETF

Direxion Daily AMZN Bull 2X Shares

Alphabet - C

GraniteShares 2x Long MSFT Daily ETF

GraniteShares 2x Long AMZN Daily ETF

Microsoft

Direxion Daily MSFT Bull 2X Shares

Direxion Semicon Bull 3X

Amazon

iShares Expanded Tech Software Sector ETF

More intense than the Internet bubble! Tech giants gamble $2 trillion on AI, with unprecedented capital intensity