Morgan Stanley Research Report Analysis: Five Major Giants' AI Spending to Reach 1.4 Trillion in 2028, Meta API Business Most Explosive
TechFlow Selected TechFlow Selected
Morgan Stanley Research Report Analysis: Five Major Giants' AI Spending to Reach 1.4 Trillion in 2028, Meta API Business Most Explosive
If previously you only considered capital expenditure and capacity planning when selecting targets, now you need to incorporate a new dimension: who has already established sales channels while building.
Navigating Web3 tides with focused insightsBy: Rita
TechFlow Guide
Morgan Stanley significantly raised its capital expenditure forecast for hyperscalers, expecting the combined capex of the top five companies (Microsoft, Google, Amazon, Meta, SpaceX) to reach approximately $1.2 trillion/$1.4 trillion in 2027/2028. By 2028, available compute capacity will grow from about 30GW in 2025 to nearly 120GW, a 4-fold increase. Morgan Stanley explicitly listed Meta as the top pick in this report, citing five-fold market options that are not fully priced in, with the API business being the most noteworthy. The Muse Spark 1.1 model is priced 30%-86% lower than peers, and every 100MW of compute capacity can leverage about $8 billion in revenue. The endgame of the compute arms race is not about who builds faster, but who sells better.
Capital Expenditure Continuously Raised, Reaching $1.4 Trillion in 2028
Morgan Stanley adopted a bottom-up cost model, calculating the construction cost per GW of data centers from two dimensions: in-rack IT hardware (chips, memory, CPU, network) and out-of-rack costs (construction materials, electrical mechanical systems). Due to rising memory prices and out-of-rack cost inflation, the cost per GW has been raised by about 20%. Coupled with GPU/ASIC supply allocation, Morgan Stanley raised the total capital expenditure of the top five hyperscalers for 2027/2028 to approximately $1.2 trillion/$1.4 trillion. Among them, Meta's capital expenditure for 2027/2028 was raised by 29%/22%, to $225 billion/$250 billion; Amazon's company-wide capital expenditure was raised by 15%/29%, to $308 billion/$318 billion; Google had previously been raised significantly.
Compute Capacity to Quadruple, AWS Holds the Largest Capacity
Morgan Stanley expects the combined available compute capacity of the top five hyperscalers to approach 120GW by 2028, a 4-fold increase from about 30GW in 2025. Among them, AWS will have the largest capacity in 2028 (35GW), followed closely by Google at 31GW, and Meta will grow from about 3.5GW at the end of 2025 to 14/21GW in 2027/2028. 55%/90% of Meta's new capacity in 2026/2027 comes from self-built facilities, with the rest from third-party leasing.
Meta Muse Spark Pricing Aggressive, API Revenue Space Huge
Morgan Stanley focused on analyzing Meta's API monetization opportunities. Meta's latest AI model Muse Spark 1.1 has opened API calls to third parties, priced at $1.25 per million input Tokens and $4.25 for output, 30%-86% lower than similar market products. Morgan Stanley built an API revenue model: every 100MW of GB300 compute capacity used for API services can generate about $8 billion in revenue and about $1.9 per share EPS increment, equivalent to about 6% upside space for 2028 EPS. On the paying user side, Morgan Stanley assumes that 25% (about 4 million) of Meta's 15 million advertisers pay about $200 per month, which can also generate about $8 billion in annual revenue. Morgan Stanley pointed out that Meta only needs to take 100MW from the total capacity of about 21GW at the end of 2028 to support this business, and the remaining capacity still has huge monetization space.
Amazon and Google: Infrastructure Beneficiaries
Amazon AWS revenue growth rate is expected to reach 40%/36% in 2027/2028, Morgan Stanley expects EPS for 2027/2028 to be $11.53/$15.05, with a target price of $330. Google has the largest compute capacity increment, adding 9GW/11GW in 2027/2028 respectively. Morgan Stanley maintains an Overweight rating on both.
TechFlow Perspective
The deciding factor in the compute arms race is shifting from "how much is built" to "how much is sold". In the past two years, the market focused on who hoarded the most GPUs; next, it depends on who can convert compute capacity into monetizable revenue streams. Meta received the highest score in this report. It is not the fastest builder, but it has the most monetization paths among the five: core advertising, subscription revenue, NeoCloud compute rental, API model calls, diffusion models, and business Agents. The aggressive pricing strategy of Muse Spark itself is a strategy to maximize compute utilization. Idling compute capacity is worthless; low-price volume running can at least cover marginal costs and build an ecosystem. If previously only capital expenditure and capacity planning were looked at to select targets, now a new dimension needs to be added: who has set up sales channels while building.
Disclaimer
This article is a compilation and interpretation by TechFlow Research of a third-party broker research report (Morgan Stanley, July 12, 2026). The ratings, target prices, earnings forecasts, and related judgments cited in the text are the views of the broker's analysts, represent only the position of their affiliated institution, do not represent the views of TechFlow Research, and do not constitute any investment advice.
The market has risks, decisions need to be independent. This article should not be used as a basis for buying or selling any securities.
Join TechFlow official community to stay tuned
Telegram:https://t.me/TechFlowDaily
X (Twitter):https://x.com/TechFlowPost
X (Twitter) EN:https://x.com/BlockFlow_News
@chaoxiangooo
