What Went Wrong With Microsoft Stock? The AI Spending Problem Behind the Selloff

Short answer: Microsoft stock fell because investors began questioning the cost, timing, and eventual returns of the company’s AI infrastructure buildout—not because Microsoft’s core business suddenly collapsed.

MSFT reached an intraday high of approximately $555.45 on July 31, 2025, then fell to roughly $349.20 in June 2026—a decline of about 37%. After Microsoft’s July 29, 2026 earnings report, the stock rebounded approximately 15.5% in one session and was around $499.99 on August 8, according to the cited market-data feeds. Prices vary slightly by feed and trading session; see Microsoft’s historical prices and Yahoo Finance’s MSFT quote page.

#	Preview	Product	Price
1		ASUS Dual Radeon RX 9060 XT 16GB GDDR6 Gaming Graphics Card	$529.00	Buy on Amazon
2		GIGABYTE GeForce RTX 5070 Ti Gaming OC 16G Graphics Card, 16GB 256-bit GDDR7, PCIe 5.0, WINDFORCE...	$1,162.49	Buy on Amazon
3		GIGABYTE GeForce RTX 5060 WINDFORCE OC 8G Graphics Card, Cooling System, 8GB 128-bit GDDR7, PCIe...	$459.99	Buy on Amazon
4		GIGABYTE Radeon RX 9070 XT Gaming OC 16G Graphics Card, PCIe 5.0, 16GB GDDR6, GV-R9070XTGAMING...	$814.28	Buy on Amazon
5		ASUS Prime Radeon RX 9070 XT 16GB GDDR6 OC Edition Gaming Graphics Card	$829.00	Buy on Amazon

The underlying company was still growing rapidly. Fiscal 2026 revenue reached $331.8 billion, Azure growth accelerated to 43% in the latest quarter, and Microsoft had more than 30 million paid Microsoft 365 Copilot seats. But capital expenditure rose to approximately $41 billion in fiscal Q4, Microsoft Cloud gross margin fell to 65%, and free cash flow was much lower than accounting earnings. The market started treating Microsoft less like a high-margin software compounder and more like an AI infrastructure operator that must spend enormous sums before investors can verify the returns.

That is an expectations and capital-allocation problem. The July 2026 quarter repaired part of the damage, but it did not settle the central question: can Microsoft turn its unprecedented AI investment into durable revenue growth, expanding cash flow, and attractive returns on capital?

#1 Best Overall

ASUS Dual Radeon RX 9060 XT 16GB GDDR6 Gaming Graphics Card

• Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
• 2.5-slot design allows for greater build compatibility while maintaining cooling performance
• 0dB technology lets you enjoy light gaming in relative silence
• Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles
• Dual ball fan bearings last up to twice as long as sleeve bearing designs

$529.00

Buy on Amazon

What went wrong with Microsoft stock, exactly?

There are four different things investors may mean when they say something went wrong with Microsoft:

The most accurate thesis is therefore:

Microsoft stock was repriced because the market became less certain that AI-related revenue would arrive quickly enough, or carry high enough margins, to justify the scale of Microsoft’s infrastructure spending.

This does not prove that Microsoft spent too much. It means investors demanded stronger evidence that the spending would produce sufficiently high returns.

The Microsoft stock selloff: a timeline

Why strong earnings did not automatically lift the stock

Public companies are not valued only on what they earned last quarter. They are valued on the expected stream of future cash flows, adjusted for risk. A company can beat revenue and earnings estimates while its stock falls if the market expected an even better result or became less confident about future returns.

Microsoft’s recent results illustrate the difference between headline growth and the quality, cost, and durability of that growth.

Microsoft’s fiscal Q2 GAAP net income also included a $7.6 billion gain related to its OpenAI investment. Excluding that impact, Microsoft reported adjusted net income of $30.9 billion, up 23%. Fiscal Q4 included a $480 million negative OpenAI investment impact and a separate $3.2 billion gain from Microsoft’s Anthropic investment. Those equity-investment gains and losses can move GAAP earnings without representing recurring revenue from Microsoft’s operating businesses.

That is why investors focused on Azure growth, capital spending, free cash flow, cloud margins, backlog quality, and forward guidance rather than simply asking whether Microsoft had beaten the consensus earnings estimate.

AI capital expenditure became the central problem

Microsoft is building data centers and buying the computing equipment required to train and run AI models. That includes GPUs, CPUs, networking equipment, power systems, cooling infrastructure, data-center construction, and leases.

In fiscal Q2 2026, Microsoft reported:

• $37.5 billion of capital expenditure.
• Approximately two-thirds of that spending directed toward short-lived assets, primarily GPUs and CPUs.
• $29.9 billion of cash paid for property and equipment.
• $35.8 billion of operating cash flow.
• $5.9 billion of free cash flow.
• Microsoft Cloud gross margin of 67%, down year over year.

By fiscal Q4, capital expenditure had risen to approximately $41 billion, while Microsoft paid $35.8 billion for property and equipment. Operating cash flow was $55.4 billion and free cash flow was $19.6 billion. The numbers remained positive, but the amount of cash consumed by the buildout was unmistakable. Microsoft’s Q2 earnings discussion and Q4 earnings discussion explain management’s view of the spending.

Microsoft was not burning cash

It would be inaccurate to say that Microsoft was losing money or burning through cash. The company remained highly profitable and generated positive free cash flow.

The more precise concern is that AI infrastructure spending consumed an unusually large portion of the cash Microsoft generated. That reduced the cash available for dividends, buybacks, debt reduction, acquisitions, or other investments, at least during the buildout phase.

Three measures should be kept separate:

• Net income is accounting profit after expenses, taxes, depreciation, and other items. It can include noncash investment gains or losses.
• Operating cash flow measures cash generated by operations before capital spending.
• Free cash flow is generally operating cash flow after capital spending. It is closer to the cash available for shareholder returns, although lease accounting and timing can complicate the comparison.

A large gap between net income and free cash flow is not automatically a sign of fraud or failure. It is common during an investment cycle. It does, however, make the return on that investment much more important.

Why investors could not simply match capex with Azure revenue

Microsoft’s AI spending is not dedicated exclusively to Azure revenue. CFO Amy Hood said investors were trying to correlate capital expenditure directly with Azure revenue, but Microsoft allocates computing capacity among several uses:

• Azure workloads purchased by external customers.
• OpenAI and other model-company workloads.
• Microsoft 365 Copilot.
• GitHub Copilot.
• Security Copilot.
• AI features in Dynamics, Teams, Windows, and other products.
• Microsoft’s own models and internal research and development.
• Replacement of aging servers, networking equipment, and other infrastructure.

That allocation can be strategically rational. A GPU may create Azure consumption revenue, improve a Microsoft 365 product, support research, or strengthen Microsoft’s competitive position without producing a separately reported revenue line immediately.

It also makes the investment case harder to model. Investors cannot reliably compare one dollar of GPU spending with one dollar of Azure revenue and conclude that the return is attractive. They must estimate how the same infrastructure contributes across several products, how long it remains useful, and how much gross profit remains after depreciation, electricity, leasing, and inference costs.

The difference between economic investment and reported capex

Microsoft’s capital-expenditure figures also require accounting context. Cash paid for property and equipment, reported capital expenditure, finance leases, and operating leases do not always move together in the same quarter.

Rank #2

GIGABYTE GeForce RTX 5070 Ti Gaming OC 16G Graphics Card, 16GB 256-bit GDDR7, PCIe 5.0, WINDFORCE Cooling System, GV-N507TGAMING OC-16GD Video Card

• Powered by the NVIDIA Blackwell architecture and DLSS 4
• Powered by GeForce RTX 5070 Ti
• Integrated with 16GB GDDR7 256bit memory interface
• PCIe 5.0
• WINDFORCE cooling system

$1,162.49

Buy on Amazon

• Finance leases can be included in reported capital-expenditure measures and bring the related asset and liability onto the balance sheet.
• Operating leases are generally treated differently and may not appear in the headline capital-expenditure number in the same way, even though Microsoft is committing economically to data-center capacity.
• Cash paid for property and equipment reflects cash timing and can differ from the reported capex measure because of leases and unpaid commitments.

Microsoft previously discussed approximately $190 billion of calendar-2026 capex and later referred to approximately $175 billion. That should not be read as an equivalent $15 billion reduction in the underlying AI investment plan. Microsoft said the change was substantially related to more future data-center leases being classified as operating leases rather than finance leases. In other words, part of the apparent reduction was accounting presentation, not necessarily less economic capacity being secured.

Microsoft also extended the estimated useful lives of data centers and office buildings from 15 to 25 years beginning in fiscal 2027. A longer useful life can reduce annual depreciation expense, but it does not automatically improve the economic return on the asset. The asset still has to generate enough revenue and gross profit over its useful life. The change also does not mean short-lived GPUs will remain economically useful for 25 years.

Azure was strong—but the market needed it to be exceptional

Azure growth was not a collapse story. It was a valuation story.

Azure and other cloud services grew 39% in fiscal Q2 2026. Microsoft guided to approximately 37% to 38% constant-currency growth for the following quarter. That was still an extraordinary growth rate, but it was modestly slower than the rate investors wanted to see while Microsoft was committing tens of billions of dollars to AI capacity.

Microsoft said customer demand exceeded available supply. The company was capacity constrained by the availability of GPUs, data centers, power, and deployment capacity. This distinction matters:

• Bullish interpretation: customers wanted more Azure capacity than Microsoft could provide, so new capacity could be monetized quickly.
• Risk interpretation: Microsoft had to spend heavily before recognizing the revenue, and supply shortages made it difficult to tell whether all future demand would remain durable and profitable.

Capacity constraints are therefore not automatically bullish. They are bullish only if Microsoft can bring capacity online at attractive returns and retain customers long enough to recover the investment. If demand weakens, if customers negotiate lower prices, or if hardware becomes obsolete quickly, the company could be left with expensive infrastructure whose expected utilization is lower than planned.

The July quarter offered better evidence. Azure growth accelerated to 43%, and Microsoft guided to approximately 45% constant-currency growth for fiscal Q1 2027. Management also said additional capacity brought online during the quarter was quickly monetized. That suggests at least some of the new infrastructure was being used rather than sitting idle.

Still, the long-term test is not just whether Azure grows. It is whether Azure revenue grows faster than the economic burden of the capacity required to produce it.

There is no single Microsoft AI revenue number

Investors sometimes talk about Microsoft’s AI revenue as though it were a single disclosed segment. It is not. The company’s AI opportunity spans multiple revenue streams with different economics:

1. Azure infrastructure: external customers pay for computing, storage, databases, and AI services.
2. Model-company consumption: OpenAI and other frontier-model companies purchase or consume Azure services.
3. Microsoft 365 Copilot: businesses pay for AI functionality layered onto Microsoft 365.
4. GitHub Copilot: developers pay for AI-assisted coding products and related usage.
5. Security Copilot: AI is incorporated into Microsoft’s security products.
6. AI features in Dynamics, Teams, Windows, and other products: the monetization may occur through higher subscription tiers, retention, upselling, or greater product usage.
7. Internal research and development: computing capacity can support Microsoft’s own models and products without creating immediate external revenue.

Azure revenue and Microsoft Cloud revenue are disclosed, but Microsoft does not report a single standardized total for all revenue directly attributable to AI. That makes claims that a precise percentage of Microsoft’s AI revenue comes from OpenAI difficult to verify from the company’s public disclosures.

OpenAI created both opportunity and concentration risk

OpenAI is an important part of Microsoft’s AI strategy, but it also made the company’s backlog and returns harder to assess.

Under the October 2025 partnership agreement filed as SEC Exhibit 99.2:

• Microsoft held an investment valued at approximately $135 billion, representing roughly 27% of OpenAI Group on an as-converted diluted basis.
• Microsoft’s intellectual-property rights were extended through 2032.
• OpenAI contracted to purchase an additional $250 billion of Azure services.
• Microsoft no longer retained a right of first refusal to be OpenAI’s compute provider.
• OpenAI could serve certain products through other cloud providers.

In fiscal Q2, Microsoft reported commercial remaining performance obligations, or RPO, of $625 billion, with approximately 45% attributed to OpenAI. That does not mean OpenAI represented 45% of Microsoft’s revenue. RPO is contracted revenue that has not yet been recognized, and the timing, consumption, pricing, and profitability of those commitments matter.

The concentration raised reasonable questions:

• Can OpenAI remain financially capable of honoring its commitments?
• How much of Azure’s AI growth is diversified across ordinary enterprises?
• Could OpenAI shift more workloads to Oracle, CoreWeave, AWS, or its own infrastructure?
• Does Microsoft earn attractive operating profits from the relationship if OpenAI remains loss-making?
• Does Microsoft retain enough control to protect the expected economics?
• Are the contracted workloads being consumed at the pace investors expect?

These are counterparty and durability questions, not evidence that the contracts are fake. It would be inaccurate to call Microsoft’s results fraudulent or a proven circular-financing scheme without authoritative evidence.

The fiscal Q4 disclosures reduced some of the concentration concern. Commercial RPO rose 84% to $678 billion, RPO excluding OpenAI rose 25%, and Microsoft said nearly 90% of full-year Microsoft Cloud revenue came from customers outside frontier-model companies. That is meaningful evidence that Microsoft’s cloud business is broader than OpenAI, although it does not eliminate the need to monitor OpenAI-related exposure.

How to read Microsoft’s backlog

RPO is useful because it shows contracted future performance, but it is not the same as current revenue, cash, or profit. A backlog can be recognized over multiple periods and may depend on customer usage, deployment schedules, pricing, and contractual terms.

Rank #3

Sale

GIGABYTE GeForce RTX 5060 WINDFORCE OC 8G Graphics Card, Cooling System, 8GB 128-bit GDDR7, PCIe 5.0, Manufactured by NVIDIA, DisplayPort & HDMI - Video Output Interface, GV-N5060WF2OC-8GD Video Card

• Powered by the NVIDIA Blackwell architecture and DLSS 4
• Powered by GeForce RTX 5060
• Integrated with 8GB GDDR7 128bit memory interface
• PCIe 5.0
• WINDFORCE cooling system

$459.99

Buy on Amazon

For each quarter, investors should ask:

• How much RPO is expected to be recognized within the next 12 months?
• How quickly is the balance converting into reported revenue?
• How much growth remains after excluding OpenAI?
• How concentrated is the balance among a small number of customers?
• Are customers consuming capacity or merely holding contractual rights?
• Will the infrastructure supporting the commitments remain useful for its full depreciation period?

The headline RPO number is encouraging, but it should not be treated as guaranteed future free cash flow.

Copilot adoption was encouraging, but adoption is not the same as profit

Copilot was a major reason investors accepted Microsoft’s AI spending. The logic was straightforward: Microsoft already had a huge enterprise software customer base, and AI could become a valuable paid layer on top of Microsoft 365, GitHub, security, Dynamics, and other products.

By fiscal Q4 2026, Microsoft reported more than 30 million paid Microsoft 365 Copilot seats, accelerating seat additions, and usage-based pricing initiatives. Microsoft 365 commercial cloud revenue grew 14% on a reported basis, or 16% on an adjusted basis. Premium offerings such as Copilot, E5, and E7 helped drive average revenue per user higher.

Those are positive signs. But a seat count alone does not prove that Copilot is a high-return business. The relevant questions are:

• How many paid seats are being added net of cancellations?
• What revenue is incremental rather than bundled into an existing E5 or E7 contract?
• What is the average revenue per paid seat?
• How much usage-based revenue is developing?
• Are enterprise customers renewing after initial pilots?
• Do customers expand Copilot across their workforces?
• How much does inference, support, and customer acquisition cost per seat?
• Does Copilot improve Microsoft 365 retention and expansion?
• Are margins improving as models, software, and Microsoft-designed chips become more efficient?

Microsoft said Microsoft Cloud gross margin declined partly because of increased Copilot usage and ongoing investment. That creates a temporary or potentially structural trade-off: Copilot can increase revenue and customer value while initially costing Microsoft more to serve.

The bullish outcome is that Copilot becomes an incremental, durable software layer with high lifetime gross profit. The bearish outcome is that seat growth looks impressive while customers resist higher prices, inference costs absorb much of the revenue, or Copilot mainly shifts existing Microsoft 365 spending instead of expanding it.

Cloud margins exposed the cost of the AI transition

Microsoft’s traditional software model benefited from attractive gross margins and relatively modest incremental distribution costs. AI infrastructure is different. It requires:

• GPU and CPU depreciation.
• Data-center construction and leases.
• Electricity, cooling, and networking.
• Model-training costs.
• Inference costs each time customers use an AI feature.
• Potential hardware obsolescence.
• Ongoing research and product investment.

Microsoft Cloud gross margin was 67% in fiscal Q2 and 65% in fiscal Q4, down year over year. The mix shift toward Azure, continuing AI infrastructure investment, and increased AI product usage all contributed to the pressure. Efficiency improvements in Azure and Microsoft 365 partly offset the decline.

The market was not necessarily demanding that margins never fall. Investors were asking whether the margin pressure was a temporary investment phase or the new economics of the business. If Azure grows quickly and margins stabilize as infrastructure is utilized, the spending may look sensible in hindsight. If revenue growth requires permanently lower margins and ever-higher capital expenditure, Microsoft’s historical software valuation becomes harder to defend.

Microsoft’s non-AI businesses made the AI question more important

Microsoft’s other businesses were not uniformly weak, but some were not accelerating enough to offset concern about the AI buildout.

In fiscal Q4 2026, More Personal Computing revenue fell 4%. Windows OEM and Devices revenue fell 7%, while Xbox content and services revenue fell 10%. Microsoft said fiscal 2027 Windows OEM and Devices revenue could decline in the high teens. It also expected on-premises server revenue to decline as customers continued shifting to cloud services.

These businesses do not explain the entire share-price decline, and Microsoft’s broader portfolio still includes Microsoft 365, LinkedIn, Dynamics, search and advertising, security, gaming, and cloud services. But the mix matters. When traditional growth is less exciting and AI infrastructure requires enormous investment, investors become more sensitive to any sign that the AI payoff could be delayed.

The key question is not whether Windows or Xbox is collapsing. It is whether Microsoft can continue growing its core subscription, business applications, advertising, and cloud franchises while AI investment absorbs capital and pressures margins. Microsoft’s annual report and 2025 Form 10-K provide the broader segment and risk disclosures.

Broader AI-stock rotation amplified Microsoft’s decline

Microsoft’s selloff occurred during a wider market debate over whether the AI infrastructure buildout was becoming excessive. Alphabet, Amazon, Meta, Microsoft, and other companies were committing enormous sums to data centers, chips, and power.

That created several market-wide concerns:

• Will AI revenue grow fast enough to justify the spending?
• Will more efficient models reduce the amount of computing required per task?
• Will customers eventually demand lower prices as model costs decline?
• Which companies will capture the profits: cloud providers, chip suppliers, model companies, or software distributors?
• Would investors prefer lower-capital-intensity software companies or businesses with clearer near-term AI monetization?

More efficient AI models create a genuine trade-off. They can increase adoption by making AI cheaper for customers, but they can also reduce the amount of infrastructure revenue generated per unit of output. That is good for AI users but not automatically good for every owner of expensive computing capacity.

The broader rotation does not explain the entire move in MSFT. It did, however, make a company-specific valuation reset more severe. Investors could rotate toward semiconductor suppliers, lower-capex software businesses, value stocks, or cloud providers perceived to offer better near-term returns.

Rank #4

Sale

GIGABYTE Radeon RX 9070 XT Gaming OC 16G Graphics Card, PCIe 5.0, 16GB GDDR6, GV-R9070XTGAMING OC-16GD Video Card

• Powered by Radeon RX 9070 XT
• WINDFORCE Cooling System
• Hawk Fan
• Server-grade Thermal Conductive Gel
• RGB Lighting

$814.28

Buy on Amazon

What the July 2026 results changed

The July 29 fiscal Q4 report addressed several of the market’s most immediate fears.

Evidence in Microsoft’s favor

• Azure accelerated: growth reached 43%, up from 39% in fiscal Q2, and management guided to approximately 45% constant-currency growth for fiscal Q1 2027.
• New capacity was monetized quickly: Microsoft said capacity brought online during the quarter was rapidly put to work.
• Free cash flow remained positive: fiscal Q4 free cash flow was $19.6 billion, and Microsoft expected to remain free-cash-flow positive in fiscal 2027.
• Backlog broadened: commercial RPO reached $678 billion, while RPO excluding OpenAI grew 25%.
• Copilot adoption strengthened: paid Microsoft 365 Copilot seats exceeded 30 million, with accelerating additions and usage-based pricing initiatives.
• Cloud growth remained substantial: Microsoft Cloud revenue reached $59.3 billion, up 27%.

The market rewarded those developments with an approximately 15.5% one-day gain, Microsoft’s strongest daily performance in nearly 18 years according to Associated Press reporting.

What did not improve enough to end the debate

• Spending remained enormous: Q4 capex was approximately $41 billion, and fiscal Q1 2027 capex was expected to exceed $50 billion.
• The apparent capex reduction was partly accounting: the move from approximately $190 billion to approximately $175 billion of calendar-2026 capex did not represent an equivalent reduction in underlying investment.
• Cloud margins remained below earlier levels: Microsoft Cloud gross margin was 65% in Q4.
• Fiscal 2027 remained an investment year: Microsoft expected capital expenditure to grow year over year, although full-year operating margin was expected to decline by less than one percentage point.
• Some traditional businesses faced pressure: Windows OEM and Devices and on-premises server revenue were expected to weaken.

The rally therefore showed that investors had become too pessimistic about near-term monetization, not that Microsoft had eliminated every long-term risk.

How to decide whether the Microsoft stock problem is fixed

Shareholders and prospective investors do not need a single dramatic prediction. They need to watch whether growth, margins, and cash generation begin moving in the right direction together.

1. Compare Azure growth with infrastructure spending

Track Azure reported growth, constant-currency growth, growth excluding unusually large contracts where disclosed, the pace of capital expenditure, capacity constraints, and the timing of data-center deployments.

More encouraging: Azure growth accelerates while capital-spending growth moderates or produces visibly more revenue and gross profit.

More concerning: capital expenditure keeps rising while Azure growth slows and Microsoft provides less evidence of utilization or returns.

2. Watch free-cash-flow conversion

Follow operating cash flow, cash paid for property and equipment, finance leases, operating lease payments, free cash flow as a percentage of net income, and free cash flow after dividends and buybacks.

More encouraging: free cash flow rises as the infrastructure cycle matures, even while Microsoft continues investing.

More concerning: accounting earnings rise for several years while free cash flow remains heavily compressed.

3. Monitor Microsoft Cloud gross margin

Microsoft Cloud gross margin is a useful high-level indicator, although it does not isolate every Azure or Copilot economic detail. Also watch commentary about GPU utilization, inference efficiency, Azure software improvements, and Microsoft’s Maia and Cobalt chips.

More encouraging: Azure growth remains high while cloud margins stabilize or expand.

More concerning: AI growth requires permanently lower cloud margins and greater capital intensity.

4. Measure Copilot economics, not just seat count

The important metrics are paid seats, net additions, average revenue per user, usage-based billing, renewal and expansion rates, incremental revenue, and gross profit after inference costs.

More encouraging: customers renew and expand, revenue per user rises, and serving costs grow more slowly than revenue.

More concerning: seat growth remains strong but paid conversion, retention, pricing power, or margins are weak.

Best Value

Sale

ASUS Prime Radeon RX 9070 XT 16GB GDDR6 OC Edition Gaming Graphics Card

• Axial-tech fans now feature a smaller fan hub that facilitates longer blades and a barrier ring that increases downward air pressure
• Phase-change GPU thermal pad helps ensure optimal heat transfer, lowering GPU temperatures for enhanced performance and reliability
• 2.5-slot design allows for greater build compatibility while maintaining cooling performance
• Dual-ball fan bearings last up to twice as long as standard conventional sleeve bearings designs
• 0dB technology lets you enjoy light gaming in relative silence

$829.00

Buy on Amazon

5. Separate OpenAI growth from diversified demand

Continue tracking RPO excluding OpenAI, Azure revenue from non-frontier-model customers where Microsoft discloses it, OpenAI’s financial capacity, its use of other clouds, and the economics of Microsoft’s equity and commercial relationship.

More encouraging: Microsoft continues producing strong growth from ordinary enterprises and industries outside frontier-model companies.

More concerning: reported AI growth slows sharply when OpenAI-related commitments are excluded.

6. Scrutinize depreciation and useful-life assumptions

Extending the estimated life of data centers and office buildings from 15 to 25 years can change the timing of depreciation. It does not prove that the assets will earn attractive returns, and it does not eliminate the shorter useful-life risk of GPUs and CPUs.

Investors should ask whether the infrastructure is likely to generate enough revenue and gross profit over its economic life—not merely whether accounting depreciation is lower in a particular year.

7. Check the health of the traditional portfolio

Watch for stabilization in Windows OEM, a return to growth in Xbox, continued growth in Dynamics and LinkedIn, resilience in search and advertising, and sustained Microsoft 365 core subscription growth independent of Copilot. Cloud migration can offset on-premises server declines, but the transition still has to produce attractive cash returns.

Important accounting and legal cautions

Microsoft’s reported earnings can be affected by the value of its investments in OpenAI and Anthropic. The fiscal Q2 OpenAI gain, the fiscal Q4 OpenAI loss, and the fiscal Q4 Anthropic gain are not the same as recurring operating revenue from Azure or Microsoft 365. Comparing GAAP net income with free cash flow without adjusting for these differences can make the business look either stronger or weaker than its underlying operations.

Shareholders also filed a proposed class action alleging that Microsoft misrepresented Azure growth, AI spending, and Copilot performance. Microsoft said the claims were without merit and that it stood by the integrity of its public statements. The allegations have not been established by a final court judgment. Readers should distinguish a complaint from proven wrongdoing; see the case docket, filed complaint, and reported coverage.

So, is Microsoft stock broken?

The evidence through August 9, 2026 does not support describing Microsoft as a broken business. Fiscal-year revenue was $331.8 billion, operating income was $155.2 billion, and net income was $133.7 billion. Azure was growing faster again, Copilot adoption was expanding, and Microsoft had positive free cash flow.

But the stock’s earlier decline was not irrational. Microsoft was priced for exceptional AI execution. Investors had already assumed that Azure would grow quickly, Copilot would monetize, OpenAI would remain a durable partner, margins would hold up, and data-center spending would generate very high returns. When capex accelerated, free cash flow weakened, cloud margins fell, and Azure growth temporarily decelerated, the market reduced the premium it was willing to pay.

The July rally means the market received better evidence on near-term demand and monetization. It does not prove that the long-term return on Microsoft’s AI infrastructure is settled.

Frequently Asked Questions

Did Microsoft miss its earnings when the stock fell?

Not in the simple sense. Microsoft continued to report strong revenue, operating income, and earnings, and it often met or exceeded headline expectations. The stock fell because investors focused on the cost and quality of growth: record AI capital expenditure, slower-than-hoped Azure growth, weaker free-cash-flow conversion, margin pressure, and uncertainty about the timing of returns.

Was Microsoft’s Azure slowdown caused by weak demand?

Microsoft said demand exceeded available supply and described the business as capacity constrained. That suggests the immediate issue was not simply weak demand. However, capacity constraints also require Microsoft to spend heavily before recognizing revenue, so investors still need evidence that the demand remains durable and profitable once new capacity comes online.

Does OpenAI account for 45% of Microsoft’s revenue?

No. The approximately 45% figure referred to OpenAI’s share of Microsoft’s $625 billion commercial remaining-performance-obligation balance in fiscal Q2 2026. RPO is contracted future performance, not current revenue. In fiscal Q4, Microsoft said RPO excluding OpenAI rose 25% and nearly 90% of full-year Microsoft Cloud revenue came from customers outside frontier-model companies.

Did Microsoft really cut its 2026 AI spending from $190 billion to $175 billion?

The reported change should not be treated as a straightforward $15 billion reduction in economic investment. Microsoft said the difference was substantially related to the classification of future data-center leases as operating leases rather than finance leases. Reported capex and the total economic commitment to infrastructure are not identical.

Has Microsoft fixed the problem after its July 2026 stock rebound?

The July results improved the near-term case: Azure grew 43%, management guided to about 45% growth for fiscal Q1 2027, RPO outside OpenAI grew, Copilot seats exceeded 30 million, and free cash flow remained positive. The longer-term issue remains unresolved until Microsoft demonstrates that AI growth can support durable free-cash-flow growth and attractive returns without permanently depressing cloud margins.

The Bottom Line

Bottom line: Microsoft stock fell because investors became less willing to pay a premium for an AI strategy requiring enormous immediate spending, declining cloud margins, and uncertain payback. The company was still growing—not collapsing—but the market wanted proof that GPUs, data centers, Copilot, and OpenAI-related commitments would produce durable, high-return cash flows. The July 2026 results supplied meaningful evidence in Microsoft’s favor, especially stronger Azure growth and broader backlog growth. The decisive metrics from here are Azure growth relative to capex, free-cash-flow conversion, Microsoft Cloud margins, Copilot retention and profitability, and RPO growth excluding OpenAI.

Quick Recap

Bestseller No. 1

ASUS Dual Radeon RX 9060 XT 16GB GDDR6 Gaming Graphics Card

0dB technology lets you enjoy light gaming in relative silence; Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles

$529.00

Bestseller No. 2

GIGABYTE GeForce RTX 5070 Ti Gaming OC 16G Graphics Card, 16GB 256-bit GDDR7, PCIe 5.0, WINDFORCE Cooling System, GV-N507TGAMING OC-16GD Video Card

Powered by the NVIDIA Blackwell architecture and DLSS 4; Powered by GeForce RTX 5070 Ti; Integrated with 16GB GDDR7 256bit memory interface

$1,162.49

SaleBestseller No. 3

GIGABYTE GeForce RTX 5060 WINDFORCE OC 8G Graphics Card, Cooling System, 8GB 128-bit GDDR7, PCIe 5.0, Manufactured by NVIDIA, DisplayPort & HDMI - Video Output Interface, GV-N5060WF2OC-8GD Video Card

Powered by the NVIDIA Blackwell architecture and DLSS 4; Powered by GeForce RTX 5060; Integrated with 8GB GDDR7 128bit memory interface

$459.99

SaleBestseller No. 4

GIGABYTE Radeon RX 9070 XT Gaming OC 16G Graphics Card, PCIe 5.0, 16GB GDDR6, GV-R9070XTGAMING OC-16GD Video Card

Powered by Radeon RX 9070 XT; WINDFORCE Cooling System; Hawk Fan; Server-grade Thermal Conductive Gel

$814.28

SaleBestseller No. 5

ASUS Prime Radeon RX 9070 XT 16GB GDDR6 OC Edition Gaming Graphics Card

0dB technology lets you enjoy light gaming in relative silence; Dual BIOS switch lets you toggle between Quiet and Performance BIOS profiles

$829.00

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.