In response to persistent market concerns over customer concentration and capital expenditures, Bernstein conducted an extreme scenario stress test on Oracle in its latest research report: what would the company be worth if AI clients like OpenAI fully withdrew their orders? According to the report, the model constructs a "near-worst-case" downside assumption: all AI contracts signed since Q4 fiscal 2025, including OpenAI's massive order, fail to convert into revenue, with clients neither fulfilling nor renewing agreements. Meanwhile, the core database, SaaS, and non-AI OCI businesses maintain normal growth. Under this extreme scenario, Bernstein estimates Oracle's valuation floor at $137 per share, representing only a 15% downside from the current level of approximately $160, providing investors with a clear margin of safety. In an optimistic scenario, assuming smooth execution, the price target could reach $313. The conclusion reveals a highly asymmetric risk-reward profile. Even if all AI-related revenue is excluded, Oracle's core enterprise business alone can support a robust valuation. Bernstein believes that market concerns over reliance on AI clients and capital expenditure pressures are already overpriced, making the current risk-reward ratio attractive. The $248 billion in lease liabilities is not an "unbearable burden." Bernstein further analyzed market worries about Oracle's massive data center lease commitments. The scenario investors fear most is that if clients default, Oracle would be forced to bear up to $248 billion in idle capacity costs. However, the report indicates this risk is significantly overstated. First, these lease contracts have terms of 15 to 19 years, with capacity rolling out in phases until fiscal 2030. The annual maximum risk exposure is actually only $13 billion to $16.5 billion, peaking only in 2030. Second, as long as the global economy does not enter an "AI winter," data center demand will remain high, and Oracle has full capability to repurpose idle space for its own use or subleasing. More importantly, even excluding generative AI, the global IaaS/PaaS market's potential size has reached $1.2 trillion to $1.4 trillion. The data center capacity consumed by traditional enterprise cloud migration alone is sufficient to absorb these leased spaces. In other words, Oracle's long-term contract structure provides an adequate buffer, and market panic over its capital commitments lacks substantive support. Hardware risks are manageable and will not be "trapped" by GPU inventory. Addressing market concerns about hardware capital expenditure risks, Bernstein points out that the actual exposure is quite limited. Procurement cycles for servers and GPUs typically occur 3 to 6 months before capacity goes live. If a client cancels a contract, Oracle can flexibly cancel or postpone orders before delivery without incurring significant penalties. Even if hardware has been delivered, most computing assets such as storage, networking, and memory equipment remain highly versatile and can be seamlessly redeployed for traditional SaaS and OCI businesses. The report further estimates that for every $1 of annual AI revenue, approximately $2 to $2.5 of computing assets are required, with an asset lifespan of 6 years. This implies that even in the event of client default, at most only 1 to 2 years of computing expenditure is at risk. Overall, the sunk risk associated with hardware capital expenditure is clearly overestimated. Strong fundamentals of the core business. In its latest model, Bernstein further isolated the AI business, clearly outlining the standalone value of Oracle's core operations. Even if AI data center-related revenue is completely excluded, the company's total revenue from traditional databases, SaaS, and OCI businesses alone could still reach $101 billion by fiscal 2030. More critically, after accounting for the interest costs on debt raised for AI infrastructure (including the recent $25 billion financing), earnings per share are still projected to reach $9.0. Compared to the average P/E ratio of 27.3x for software peers like Microsoft, SAP, and Salesforce, this earnings expectation corresponds to a valuation of $137 per share. In other words, even if the AI business were to go to zero, Oracle's core operations alone could support approximately 85% of the current share price, demonstrating the solid cash-generating capability of its foundational business. Bernstein believes that Oracle's current stock price has already over-discounted pessimistic expectations, and the risk-reward ratio is entering a highly attractive range. With the recent completion of approximately $25 billion in debt and equivalent equity financing, funding bottlenecks for data center construction have been substantially resolved. The report emphasizes that as long as the global economy avoids an extreme "AI winter" scenario, the current valuation level offers investors ample margin of safety and asymmetric upside potential.