In a sell-side research report, Morgan Stanley raised its estimate of the capital expenditure of major hyperscale cloud providers, projecting that the total capital expenditure of the top five platforms would reach $1.2 trillion and $1.4 trillion in 2027 and 2028, respectively, and continued to rank META as the top choice for the AI Internet with a price target of $775.

These figures are part of the research report model and do not equate to the company's official guidance. The publicly available Morgan Stanley material has indicated that by 2028, global AI-related infrastructure investment will approach $3 trillion, with data center capital expenditures around $2.9 trillion. The $1.4 trillion item for the top five platforms is more of a breakdown estimate by the sell-side regarding major cloud and internet platforms.

The most significant change in this report is the continued elevation of AI infrastructure spending. By 2028, the available compute capacity of major platforms is projected to be close to 120GW in the model, about 4 times the 30GW in 2025. The construction cost per GW has also been increased, with new-generation platforms like GB200, GB300, Vera Rubin requiring more memory, power, racks, and engineering investment.

For investors, the question has shifted from "Will AI giants spend money" to "How long will it take for this money to turn into revenue." META is positioned at the forefront due to facing higher AI capital expenditure pressure while having more direct monetization channels such as advertising, consumer apps, model APIs, and subscription tools.

$1.4 Trillion Expenditure Banking on 120GW of Compute Power

The report has raised the expected capital expenditure of the top five major hyperscale cloud providers for 2027 and 2028 by 9% and 10%, to $1.2 trillion and $1.4 trillion, respectively. This coverage includes Amazon, Google, Microsoft, META, and SPCX-related AI infrastructure spending.

Capacity expansion is one of the main reasons for the expenditure increase. In this model, the available compute capacity of the major platforms is projected to increase from around 30GW in 2025 to nearly 120GW by 2028. Amazon's total capacity is estimated to be around 35GW by 2028, Google is adding the most capacity in 2027 and 2028, while META is expected to increase from around 3.5GW at the end of 2025 to 14GW in 2027 and 21GW in 2028.

The capital expenditure estimates for META need to consider methodological differences. In the research report model, META's capital expenditure for 2027 and 2028 has been revised up to $225 billion and $250 billion, respectively. Some publicly available reports mention a Morgan Stanley estimate for META totaling around $380 billion for 2027 to 2028, which may involve different methodologies such as total capital expenditure, AI infrastructure, total spend, or off-balance sheet financing.

These differences do not change the overall narrative: AI data center spending continues to weigh on free cash flow, depreciation, and short-term EPS, and will also determine whether future cloud, advertising, search, API, and enterprise tool revenues can be realized. Those who can convert more computing power into revenue-generating products will have an easier time justifying today's capital expenditures.

Higher Cost per GW, Memory and Power Infrastructure Raise the Bar

The increase in spending is not only due to "building more data centers" but also because of "higher cost per GW."

In the bottom-up cost model provided in the research report, the construction cost per GW for GB200 is approximately $35 billion, a 16% increase from previous assumptions. For GB300, it is around $39 billion, a 19% increase. For Vera Rubin, it is about $49 billion, a 20% increase. Google TPU v7 is approximately $27 billion, and Amazon Trainium3 is about $21 billion.

The cost pressures mainly stem from two areas. The memory share in high-end AI systems continues to rise, and external data center costs such as power, land, cooling, distribution, and engineering construction are also increasing. The report assumes these related costs have increased from around $10 million/MW to approximately $11 million to $19 million/MW.

This is also why the spending curve of AI giants is unlikely to decrease significantly in the short term. While chip supply improvements can alleviate some pressure, challenges such as power access, racks, construction, skilled labor, and local approvals will continue to lengthen the construction cycle. Some project timelines may be extended to around three years, and the larger the capital expenditure, the more the revenue side will need to demonstrate returns more quickly.

The Focus of META Shifts to AI Monetization

META is considered a leader, with a key focus on AI revenue options more consolidated than most internet companies.

The research report breaks down META's potential upside into directions such as Meta AI Search, new cloud services, API revenue, subscription tools, and advertising upgrades, which could contribute approximately $10 to EPS by 2028. Under the base scenario, META's EPS in 2028 is $33.41. If some options are cashed in, there is further room for EPS to rise.

This calculation is not fully consistent with some publicly reported estimates mentioning "four products or catalysts" and "an increase of $1 to $3 in EPS by 2028," and is more suitable as a scenario calculation from this research report. What can actually be reflected on the financial statements depends on product adoption rates, pricing power, and computing power utilization.

APIs are the most straightforward entry point. Meta announced on July 9th the public preview of the Meta Model API. Third-party information from price tracking firm Artificial Analysis shows that Muse Spark 1.1 API input and output prices are $1.25 and $4.25 per million tokens, respectively, lower than some leading competitors.

The research model further assumes that for every 100MW of GB300 capacity used for APIs, equivalent to about 53,300 GPU units with a 75% utilization rate, it can generate approximately $8.59 billion in revenue, $640 million in incremental EBIT, and bring about a $1.91 EPS increment by 2028. This calculation relies on high utilization rates and sustained demand, as low prices can only help onboard customers but do not guarantee profitability on their own.

Subscription tools are also a potential entry point. The model assumes that out of META's 15 million advertisers, 25% will pay around $200 per month for business agents, coding assistants, and other tools, contributing approximately $8 billion in revenue and around $2 to 2028 EPS. Whether advertisers are willing to continue paying will ultimately depend on whether these tools can bring higher conversions, lower production costs, or stronger automation capabilities.

Amazon and Google Benefit, Revenue Validation Still to Catch Up

Amazon and Google are also key players in this round of capital expenditure increase, but they are more like background references in this narrative.

On the Amazon side, the research report raised the AWS revenue growth outlook, expecting growth of 40% and 36% in 2027 and 2028, respectively. It also estimated that AWS's backlog increased by about $110 billion to around $475 billion in the second quarter. Since Amazon has not yet released the corresponding official second-quarter financial report, this backlog number should be viewed as a seller forecast. What has been officially confirmed is that AWS's sales in the first quarter of 2026 increased by 28% year-on-year, OpenAI added a $100 billion multi-year commitment, and cash capital expenditure continues to rise.

Google's strength lies in the Gemini model, TPU, and the full-stack capability of its cloud business. The research report's model shows that Google will add the most capacity in 2027 and 2028 among major platforms. Short-term pressure may lie in computing resources still potentially constraining product launch, especially when search, cloud services, and model APIs are all competing for computing power simultaneously.

These clues point to the same real issue: AI spending has entered the trillion-dollar level, and the market will increasingly directly ask "how much revenue does each dollar of capital expenditure bring." Cloud services, AI search, APIs, advertising tools, and enterprise subscriptions will all become entry points for validating expenditure returns.

Huge Expenditures Still Need to Navigate Through Power, Approvals, and Real Demand

This round of capital expenditure increase has clear boundaries.

The first constraint is supply. Chips, HBM memory, racks, power access, and skilled labor will all affect the pace of construction. From planning to commissioning an AI data center, it needs to navigate through local approvals, grid upgrades, and construction cycles, and cannot be linearly implemented based on model assumptions.

The second constraint is political and regulatory. Large data centers' use of electricity, water resources, and land may face local resistance. Energy policy and local approval processes may also change around the 2026 U.S. midterm elections and the November 2028 presidential election.

The third constraint is demand. META's API, subscription, and advertising upgrades are still on an upward trend, and revenue realization requires real customer payments and continued usage. Price competitiveness with competitors favors customer onboarding, but long-term profitability depends on usage, gross margins, and tool ROI.

The $1.4 trillion capital expenditure paints a picture of a high-cost growth curve. Giants are pre-securing AI computing power in advance, and the market will continue to question when this computing power will translate into revenue and profit. META's $775 price target is built on the gradual realization of AI monetization, and the most challenging step is turning the EPS growth in the model into cash flow in the financial statements.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia