TL;DR

• Goldman Sachs maintains a buy rating on Microsoft and a $610 price target, implying about 59% upside based on the July 9 stock price.
• Azure growth remains the main line of financial reports. Goldman Sachs predicts a growth rate of 40%-41% in the fourth quarter, which is higher than the company’s previous guidance.
• Higher capital expenditures will amplify the return controversy, with Copilot charges, Maia chips and new capacity releases still to be realized.

Goldman Sachs maintained a buy rating ahead of Microsoft's fourth-quarter earnings report on July 29 and gave it a 12-month price target of $610, while raising its medium- and long-term capital expenditure forecasts. For investors, the focus of the financial report is not whether Microsoft is the winner of AI, but whether Azure can maintain high growth as computing power continues to increase, and allow higher data center, chip and power investment to be converted into revenue instead of dragging down free cash flow and profit margins.

With a target price of US$610, Azure will continue to exceed expectations

Quotations show that as of July 9 UTC, Microsoft’s stock price was approximately US$383.34. At this price, the $610 price target corresponds to potential upside of approximately 59.1%.

This set of calculations is based on several conditions: cloud demand maintains high growth, new data center capacity comes online as planned, Microsoft's internal AI research and development and external customer computing power allocation do not squeeze each other, and AI products such as Copilot begin to contribute clearer revenue and profits.

The first thing to be targeted in the financial report is Azure.

Microsoft's official FY26 Q3 conference call showed that revenue from Azure and other cloud services increased by 40% year-on-year and 39% at constant exchange rates. The company had previously given guidance for FY26 Q4 of 39%-40% growth at constant currencies, saying customer demand still exceeded available capacity.

Goldman Sachs reported that Azure's year-over-year growth at constant exchange rates in the fourth quarter is expected to reach 40%-41%, and guidance for the next quarter may also remain at 40%-41%. This forecast is slightly higher than the company's previous guidance, but market expectations are already not low. If Microsoft only delivers cloud growth that is in line with high expectations, the stock price may not continue to pay for higher AI investments.

Microsoft also needs to explain where the growth is coming from. It may be the release of new data center capacity, it may be the continued expansion of enterprise AI demand, or it may be smoother computing power scheduling between internal applications and external customers.

In the past few quarters, the constraint of Microsoft's AI business has not been lack of demand, but tight supply. Azure not only serves external customers such as OpenAI, but also supports Microsoft's internal Copilot, MAI model development and first-party applications. When computing power is tight, cloud growth will be limited by delivery capabilities. When capacity is released too slowly, capital expenditures will first be reflected in cash flow and depreciation pressure.

Microsoft's FY26 computing capacity expenditures are split by purpose and external/internal computing power allocation. AI computing, MAI, Copilot, etc. account for a high proportion, and internal computing power investment has stabilized after rising in the past 12 months. This is the key to judging whether Azure can support customer needs and internal AI research and development at the same time.

Capital expenditures continue to rise, and the AI computing power competition has not cooled down

Microsoft has given a signal of higher investment. FY26 Q3 capital expenditures were US$31.9 billion, and the company guided that Q4 capital expenditures will exceed US$40 billion, and expects capital expenditures in calendar year 2026 to be approximately US$190 billion, of which approximately US$25 billion will come from higher component prices.

Goldman Sachs reported that Microsoft's capital expenditure forecast for fiscal year 2028-2030 has been raised by about 10%. According to the report's calculations, some of the adjusted annual capital expenditure assumptions are higher than the consensus market expectations, reflecting a more radical judgment on Microsoft's future computing power investment.

This is not a choice made by Microsoft as a company. Guidelines from chip manufacturers such as Nvidia, Broadcom, and AMD, as well as capital actions by cloud and Internet giants such as Google and Meta, all show that the demand for AI computing power has not cooled down significantly. Hyperscale cloud vendors are still preparing to expand data centers, chips and power resources in the coming years.

For Microsoft, high investment has two sides.

On the one hand, the Azure and AI product cycles remain valuation supports. Goldman Sachs reports that Microsoft's computing power capacity is likely to expand to about 40GW by mid-2030. On the other hand, the higher the capital expenditure, the more investors will ask whether the new computing power can be converted into cloud revenue, AI subscriptions and higher gross profit business, rather than just bringing heavier depreciation and cash flow pressure.

The Goldman Sachs report also predicts Microsoft's FY26 revenue will be US$329.4 billion and EPS will be US$16.75, and FY27 revenue will be US$387.1 billion and EPS will be US$19.32. The implicit premise of this set of forecasts is that AI investment will not only boost revenue, but will not continue to slow down the rate of profit release.

Hyperscaler cloud vendors’ 2026/2027 capital expenditure street expectations. Since January, capital expenditure expectations of AMZN, META, GOOGL, MSFT, and ORCL have all been significantly raised, with MSFT expected to increase by 55% in 2027.

Copilot needs to be charged, and Maia needs to reduce GPU dependence

Whether Microsoft’s AI investment can go through will ultimately depend on two factors: the commercialization of Copilot, and the maturity of self-research and alternative chip supply.

Copilot's logic is relatively clear. The increase in usage will be beneficial to the expansion of software revenue in the long term and may also have the opportunity to improve the profit structure. But the short-term problem is that usage alone does not equal revenue realization.

Microsoft revealed in FY26 Q3 that the number of M365 Copilot paid seats has exceeded 20 million. GitHub Copilot is also moving toward more usage and value pricing. The company has also introduced fair usage terms for high-usage scenarios, trying to more tightly bind higher inference costs to the payment mechanism.

What the market needs to watch is not only the continued increase in the number of seats, but also user engagement, willingness to renew, and actual paid expansion on the enterprise side. If Copilot’s user experience and commercialization pace cannot be improved simultaneously, the realization of high gross profit margins for AI software will be delayed.

Chip and supply chain are another line. Microsoft's self-developed AI chip Maia is still in the catching-up stage, and its maturity lags behind some of its peers. Maia 300 improvements, AMD's production progress as a secondary source, and memory procurement costs will all impact Microsoft's ability to reduce its reliance on the external GPU supply chain.

The company has also previously mentioned that new supply needs to be balanced between Azure, first-party applications, research and development, and server replacement. If the new supply goes smoothly, Microsoft can deliver more computing power to external Azure customers while continuing to invest in internal AI research and development. If the release is uneven, there will still be a squeeze between Azure growth, internal model training, and Copilot inference needs.

Xbox restructuring is just the margin of valuation

In addition to the main line of AI, the Goldman Sachs report also used the SOTP method to estimate the value of Microsoft's gaming business at approximately US$30 billion.

On July 6, Microsoft announced the restructuring of its Xbox business. Multiple media reports show that Microsoft is laying off about 4,800 people, of which about 1,600 people at Xbox will be laid off immediately, and about 3,200 people will be laid off in FY27. Four studios, Compulsion, Double Fine, Ninja Theory, and Undead Labs, have left the Xbox management system, and the company has also reportedly streamlined some management.

This part is more like business structure adjustment, not the main line of financial reporting transactions. Microsoft's gaming business is still valuable, and its restructuring also shows that the company is cleaning up inefficient assets and shrinking some non-core investments. However, it is difficult to replace the return on capital expenditures of Azure, Copilot and AI in the short term, becoming the main factor explaining the direction of the stock price.

According to the SOTP valuation reported by Goldman Sachs, Intelligent Cloud is still the largest contributor to Microsoft's enterprise value. M365's commercial and consumer businesses have an implied enterprise value of approximately $492 billion, corresponding to approximately 4x EV/sales or 6x GAAP EBIT in 2027, incorporating certain disintermediation risk assumptions.

Whether the $610 can be cashed out depends on three things

The direction given by this financial report outlook is still optimistic: Microsoft is in a favorable position in the AI computing power, Copilot and agent orchestration layers, and has the opportunity to continue to benefit from the AI product cycle. But whether the $610 price target can be realized depends on whether the earnings report and conference call can provide more verifiable progress.

Azure needs to continue to deliver high growth and explain whether it can support external customer demand after new capacity comes online. If growth is simply in line with already high market expectations, higher capital spending could become a point of contention.

Maia 300 and AMD secondary sources need to provide clearer progress. Supply chain constraints, rising memory costs, and insufficient chip maturity will all affect the unit economics of Microsoft's AI investment.

Copilot must prove its true charging ability. More than 20 million paid seats are just the starting point. Enterprise-side paid expansion, usage billing and user feedback will determine whether it can transform from an AI entrance to a source of profit.

The point of Microsoft’s financial report is not whether AI investment will continue, but whether higher investment can quickly turn into Azure growth, AI software revenue and sustainable profit margins. If this evidence remains insufficient, the controversy over capital expenditure returns will continue to weigh on stock prices.