Bank of America senior analyst Vivek Arya recently described the sell-off in the software sector as a "logical contradiction." On one hand, supercomputing cloud providers and AI computing firms are being sold off due to massive capital expenditures, with the market believing the investments will not yield returns. On the other hand, the market claims AI will disrupt SaaS and replace human labor. Logically, these two views cannot coexist. If AI is the future, then heavy investment in computing power is the correct choice, and investors should buy computing stocks and sell SaaS stocks. If AI is far less important than the market believes, then computing stocks should be sold and traditional SaaS stocks should be bought. However, the simultaneous sharp decline in both categories suggests the market does not understand AI and is merely reacting blindly.
Microsoft (MSFT.US), a core member of the "Magnificent Seven" and a central figure in the AI revolution, has become a prime example of market mispricing. The sell-off in Microsoft stock due to high capital expenditures implies investors doubt the rationale behind AI computing infrastructure investment. Simultaneously, they fear Microsoft's software business could be disrupted by AI, believing companies will stop purchasing Microsoft's "seat licenses" if AI replaces human workers. Even non-experts can see the absurdity in this logic. Clearly, the current market pricing of Microsoft has become severely disconnected from its fundamentals. This irrational volatility, however, presents a rare entry opportunity for investors. Regardless of which market view ultimately proves correct, Microsoft is well-positioned to maintain its footing.
The market's primary concern regarding Microsoft stems from its substantial quarterly capital expenditure of $37.5 billion and plans for continued increased investment. Widespread worry exists that hyperscale cloud providers will face a computing power oversupply, and that Microsoft's capital spending lacks a clear correlation with Azure revenue growth, making returns on investment unverifiable. This concern, however, fundamentally misunderstands the construction cycle of computing infrastructure. According to information from Microsoft's latest earnings call, approximately two-thirds of the quarter's capital expenditure was allocated to "short-cycle assets" like GPUs and CPUs, primarily to meet persistently high customer demand exceeding supply. The company explicitly stated that Azure customer demand consistently outpaces supply, necessitating ongoing investment to balance new supply with rapidly growing Azure demand, compute requirements for Copilot products, support for R&D innovation, and the replacement of aging servers and network equipment.
More importantly, Microsoft's capital expenditure is not solely for purchasing chips from Nvidia (NVDA.US) and AMD (AMD.US); the company is accelerating vertical integration to build its own computing moat. During the Q2 earnings call, CEO Satya Nadella repeatedly emphasized the deployment of the self-developed Maia 200 AI accelerator. Nadella stated, "At the chip level, we have Nvidia, AMD, and our own Maia chip, enabling optimal cluster performance, cost, and supply across multiple hardware generations. This week we deployed the Maia 200 accelerator, delivering over 10 petaFLOPS at FP4 precision and optimizing total cost of ownership by more than 30% compared to the latest hardware. We will begin scaling its use for inference and synthetic data generation for the Super Intelligence team, as well as for inference services for Copilot and Foundry." For non-technical audiences, metrics like FP4 precision and 10 petaFLOPS may be insignificant, but a 30% reduction in total cost of ownership is a substantial benefit. Microsoft is also advancing its self-developed Cobalt 200 CPU, which offers over 50% better performance than the previous generation. This is a key positive: the company is gradually reducing its reliance on Nvidia, avoiding premium payments for all hardware, which should lead to long-term margin recovery—margin contraction being a core bearish argument against Microsoft.
Microsoft's gross margin declined to 68% this quarter, seemingly under pressure, but the core reason is the company's current investment phase: high-cost assets begin depreciating immediately upon deployment. As Microsoft gradually replaces Nvidia and AMD chips with its own Maia and Cobalt chips, the cost per inference (each user query to Copilot) will decrease significantly, driving a margin rebound. As the penetration of self-developed chips increases, efficiency gains are expected to outpace depreciation growth, which is the logic behind management's forecast of a slight operating margin improvement for the full year 2026.
Beyond capital expenditure, OpenAI is another major point of market contention regarding Microsoft. Recently, Microsoft recorded a $10 billion GAAP accounting gain from an OpenAI capital restructuring, but this is merely an accounting treatment; the underlying business logic is more complex. Approximately 45% of Microsoft's remaining performance obligation comes from selling computing power to OpenAI. On one hand, this is positive: OpenAI is a major client, and its recent funding reduces default risk. On the other hand, it introduces significant customer concentration risk—Microsoft's computing infrastructure build-out is heavily tied to OpenAI's success, and it remains unclear how OpenAI will achieve profitability to cover its massive investments. Recent news involving Anthropic complicates the situation: Anthropic lost a government contract, which OpenAI secured. Theoretically, as OpenAI's cloud provider, Microsoft should benefit, but the market reaction has been negative. The concern is that if OpenAI continues to grow, Microsoft only benefits from infrastructure sales, potentially weakening the long-term competitive advantage of its own software business. The market even worries that Microsoft's own Copilot tools, which increasingly integrate third-party models, might be perceived merely as a gateway to external technology rather than genuine innovative products, causing Microsoft to lose competitiveness in software.
However, it is crucial to recognize that Microsoft's fundamental strength has never relied on individual users but on enterprise clients. As long as enterprises continue to choose Microsoft's products and services, the company's fundamentals will not be fundamentally shaken.
**The Irreplaceable Foundation: Microsoft's Perfect Hedge**
Investors should not be bearish on AI, especially for giants with their own cloud services. "Excess computing capacity" should not be a primary concern. Even if not sold to AI clients, computing power can be utilized in countless other scenarios or for internal use. Microsoft added 1 gigawatt of computing capacity within three months and is building "AI factories" in locations like Atlanta and Wisconsin, connecting multiple data centers with high-speed networks to form massive, integrated computing organisms. Microsoft Copilot has reached 15 million paid seats, with seat growth soaring 160%—although this is still a small fraction of the total user base, the growth momentum is strong. Furthermore, like other major tech giants, Microsoft's computing capacity remains in a state of undersupply, necessitating continuous expansion.
Analyzing through Bank of America's "logical contradiction" lens, Microsoft represents a highly valuable investment target, even acting as a hedge. If AI proves highly successful and replaces human labor, Microsoft wins because it controls Azure's computing infrastructure and provides the intelligent tools enterprises need for automation. If AI only delivers moderate efficiency improvements, Microsoft still wins because it boasts 450 million Microsoft 365 commercial seats, with extremely high enterprise switching costs, ensuring a stable core business. The current market sell-off due to high capital expenditure essentially offers investors a prime opportunity: to acquire the world's leading AI infrastructure assets at a valuation priced for the pessimistic scenario of a "dying software business," while in reality, there are no signs of Microsoft's software business disappearing.
**Valuation Analysis**
Market over-pessimism has pushed Microsoft's valuation into a highly attractive range. First, consider the company's operational strength: Microsoft's remaining performance obligation stands at $625 billion, a 110% year-over-year increase. This figure alone substantiates Microsoft's core value and the necessity of its computing investments. Approximately $281 billion of this is linked to OpenAI, meaning over $344 billion comes from diversified long-term contracts, indicating a healthy structure.
Regarding valuation, Microsoft's current forward P/E ratio is only 25.53x, significantly below its long-term historical average of 33.21x. This places Microsoft's valuation in a low range not seen since 2022, and such levels have not persisted for long in the past. Essentially, the current stock price only reflects extreme skepticism, leaving no room for the company's operational stability, the long-term potential of the AI industry, or the efficiency gains from self-developed chips and high-density data centers. Once Microsoft's in-house chips achieve scale, driving down costs and repairing margins, the current price will likely be seen as a significant undervaluation.
**Conclusion**
The current market views Microsoft's substantial capital expenditure as a burden, but data suggests these investments will ultimately build a defensive moat. By aggressively pursuing self-developed chips and computing infrastructure, Microsoft is essentially locking in computing costs for the next 5-10 years. A P/E ratio of 25.53x completely ignores the core fact that Microsoft's computing capacity is in a state of undersupply. One side of the market will inevitably misjudge the future of the software industry, but based on the scale of Azure's backlog and historical returns on invested capital, the misjudged party is unlikely to be Microsoft.
This does not imply an absence of risk. For instance, even after its funding round, OpenAI's development remains highly uncertain. Copilot's commercial adoption may fall short of expectations, with user penetration slowing or reversing. The company's revenue growth, currently at 17% for the quarter, could decelerate further to 15%-16%, likely triggering a negative market reaction. Fierce competition from players like Google (GOOGL.US) and Amazon (AMZN.US) persists. However, even after fully considering these risks, Microsoft's risk-reward profile at the current valuation level remains highly attractive.