UBS: Stable Cloud Spending Fuels AI Momentum Across AWS Azure Google Cloud

UBS’s field checks at Oracle AI World suggest corporate cloud spending has moved from hand-wringing to pragmatism: budgets are broadly stable and healthy, AI-driven GPU demand remains strong, and the three hyperscalers — Amazon (AWS), Google Cloud, and Microsoft (Azure) — are positioned to capture the lion’s share of near‑term cloud dollars, albeit with different upside and execution risk profiles.

Background / Overview​

The conventional worry of the past 18 months — will enterprises cut cloud budgets? — has receded into a more nuanced reality: CIOs are holding 2025 IT budgets flat in many cases, but they are not cutting projects; instead, they are prioritizing AI, data modernization and longer-term capacity commitments. UBS analyst Karl Keirstead conveyed that message after speaking with customers and partners at Oracle’s AI World, noting that not a single Fortune 500 contact planned further cuts or delays in cloud spending. That empirical color underpins UBS’s conclusion that cloud demand is “stable, healthy” rather than fragile.
This is not wishful thinking. Multiple field signals point to materially different buying behavior compared with earlier cycles:

• Organizations are moving from pilots to contracted capacity for AI training and inference.
• Buyers increasingly prefer reserved, multi‑year capacity to avoid accelerator scarcity.
• Demand for cloud GPUs and managed AI services remains robust among model providers and startups.

Those shifts change the revenue profile for hyperscalers — increasing booked commitments and RPO (remaining performance obligations) while pushing recognition and margin dynamics into multi‑quarter tradeoffs.

---

What UBS Actually Heard at Oracle AI World​

Key takeaways (concise)​

• Tone: More positive than three months prior; a “thaw” in budget conservatism was reported by several partners.
• Budget posture: Many large enterprises expect flat IT budgets for 2025 — but none plan additional cuts or delays. Budget allocations are being re‑prioritized toward AI and data modernization.
• AI GPU demand: Sustained demand for GPU/accelerator capacity for both training and inference from AI model suppliers and startups; no significant slowdown expected in Q3 2025.
• Vendor dynamics: Partners reported Azure acceleration in Q3 (and expected further acceleration in Q4), while AWS was described as “slightly below expectations” in Q3 but stable outlook for Q4. Google Cloud showed upside potential due to strong ML tooling and enterprise traction.

These field signals are granular and actionable: they come from partner and customer conversations, not merely from headline bookings. That adds credibility to UBS’s measured optimism — but it also requires careful parsing because intent and contract conversion are not identical things.

---

How This Helps Explain the Hyperscalers’ Differing Profiles​

AI workloads and enterprise buying behavior have created asymmetric advantages and risks among the big three cloud vendors. UBS’s read — corroborated in market coverage — breaks down into three pragmatic strengths:

1) Amazon Web Services (AWS): scale and breadth​

• Strengths: unmatched global footprint, the broadest service catalogue, and in‑house silicon initiatives (Trainium/Inferentia/Graviton) that help defend TCO.
• Near‑term reality: AWS remains the revenue engine with enormous absolute scale, but its relative growth upside may be more constrained against tough comparables. UBS suggests AWS has limited upside relative to Wall Street’s expectations.

2) Microsoft Azure: monetization through integration​

• Strengths: tight integration of cloud infrastructure with productivity and enterprise software (Microsoft 365, Dynamics, Power Platform), allowing AI features (Copilot, etc.) to be monetized by seat or workflow.
• Near‑term reality: Azure is showing sequential acceleration in partner reports, and Microsoft’s ability to upsell AI across an installed base provides durable monetization leverage. That gives Azure a lower execution risk profile for translating AI interest into recurring revenue.

3) Google Cloud: tooling and ML engineering momentum​

• Strengths: deep ML/data tooling (Vertex AI, BigQuery), TPU/GPU investments, and a product stack that appeals to data‑centric engineering teams.
• Near‑term reality: Google Cloud is the fastest‑growing major cloud business in recent quarters and has upside potential to surprise investors if enterprise sales continue to improve. UBS flagged Google as having more upside than AWS in the near term.

These distinctions are important for investors and CIOs because they imply different tradeoffs: AWS offers scale and margin durability; Microsoft offers integrated monetization and lower churn; Google offers growth and product‑led wins among data/ML teams.

---

The AI Tailwind and Why It Changes Cloud Economics​

AI workloads shift three core economics in favor of hyperscalers, but they also introduce new execution risks.

• Scale of compute demand: Training large models consumes orders of magnitude more GPU/accelerator capacity than typical enterprise workloads; that creates capital intensity and long lead times.
• Contracting behavior: Organizations prefer reserved, long‑dated capacity to avoid spot shortages — creating backlog and RPO dynamics that boost visibility but delay revenue recognition.
• New monetization vectors: Managed model hosting, inference credits, and integrated data‑to‑model services often carry higher gross margins than raw compute and can lift overall cloud profitability once scale and utilization improve.

That combination explains why UBS sees “stable, healthy” spending: cloud becomes the path of least friction for scaling AI projects. But the tailwind is conditional — hyperscalers must convert bookings into recognized revenue, secure accelerated hardware supply, and manage capex vs. margin tradeoffs.

---

Capex, Accelerators and the Execution Risk​

UBS and other industry trackers repeatedly highlight the operational friction points that could blunt cloud revenue growth even with strong demand:

• Capex and depreciation pressure: Massive data‑center builds and accelerator purchases push near‑term depreciation, which can compress gross margins until utilization lifts.
• Accelerator supply constraints: Dependence on a small set of accelerator vendors (e.g., NVIDIA and select silicon partners) — combined with geopolitical export controls — creates chokepoints for converting backlog into revenue.
• Permitting, power, and construction delays: Data center rollouts face local approval, power availability, and supply‑chain constraints that can slow delivery timelines and undermine revenue conversion assumptions.

Investors should treat large RPO/backlog figures with caution until conversion cadence and named contract confirmations (or at least predictable conversion schedules) are visible in corporate reporting.

---

Market Evidence and Independent Checks​

UBS’s conclusions are supported by both market performance and independent reporting:

• Public results in mid‑2025 show Google Cloud growing faster (reported mid‑2025 quarter revenue in the low‑teens to 30%‑plus on a YoY basis), Microsoft’s Intelligent Cloud producing large, high‑teens to low‑30s growth depending on metric, and AWS remaining the largest single revenue contributor with lower, steady growth rates. Those public numbers corroborate the field color UBS reported.
• Financial press coverage ahead of earnings week echoes UBS’s theme: analysts and news services note that AI‑driven cloud demand is underpinning hyperscaler revenue trajectories even as investors watch capex and margin math.

Where UBS adds value is in the ground‑level voice of partners and customers at a focused event (Oracle AI World). That qualitative input helps explain the why behind the reported numbers — that enterprises are actively converting pilots into production with reserved capacity commitments.

---

For Investors: Tactical Implications and Positioning​

UBS’s messaging suggests a practical, differentiated allocation framework rather than an all‑in or all‑out call:

• Microsoft (Azure): Defensive growth + monetization — attractive for investors wanting AI exposure with lower dependence on raw capacity expansion because Microsoft can monetize AI across its software franchises.
• Google Cloud: High‑beta growth — offers the largest upside if enterprise sales motions and capex execution continue to scale profitably. Good for investors willing to accept operational capex risk in exchange for faster growth.
• AWS (Amazon): Scale and durability — less upside surprise but a large, cash‑generating franchise that underpins corporate profitability. Position size should reflect slower growth comparables.

Practical portfolio guidance:

1. Size positions according to risk appetite: Microsoft for income and integrated monetization; Google for growth; AWS for durability.
2. Watch capex cadence, RPO conversion, and accelerator purchase disclosures — these are the execution metrics that validate UBS’s thesis.
3. Add selective exposure to specialist data‑cloud players (Snowflake, Databricks‑style plays) that benefit from data‑software demand driven by AI.

---

For CIOs and Procurement Leaders: Practical Checklist​

UBS’s field checks contained specific operational advice for buyers — practical because many enterprise buyers are confronting the same capacity and governance risks seen by UBS:

• Negotiate SLA and capacity roadmaps tied to accelerator availability.
• Favor reserved capacity for training and managed inference for latency‑sensitive production.
• Negotiate portability and egress protections — containerization, model formats and clear data egress terms reduce vendor lock‑in.
• Treat large vendor commitments as operational delivery projects with milestones, not as frictionless revenue.

These steps reduce the operational risk of deploying mission‑critical AI workloads and help ensure capacity promises translate into production timetables.

---

What to Watch Next (Earnings Calendar and Near‑Term Catalysts)​

UBS delivered this field color just ahead of a major tech earnings week — a period when investors will look for signs that backlog and AI monetization are converting to revenue and margin improvement.

• Microsoft: Microsoft announced it will publish fiscal Q1 FY2026 results after market close on Wednesday, October 29, 2025, with an earnings webcast scheduled. That timing makes it a key event to watch for Azure commentary and capex disclosure.
• Alphabet (Google): Major earnings calendars list Alphabet’s Q3 2025 results on October 29, 2025 after the market close — another primary read on Google Cloud momentum and capex guidance.
• Amazon: Amazon is widely scheduled to release Q3 2025 results after market close on October 30, 2025, giving investors the final hyperscaler data point for the week.

Expect investors to parse these reports for:

• Cloud revenue growth rates and year‑over‑year comps.
• Comments on AI workloads, GPU supply and managed AI service adoption.
• Capex guidance and depreciation outlook.
• RPO/backlog metrics and named customer disclosures that substantiate UBS’s field signals.

---

Balanced Critique — Where UBS’s Read Is Strong and Where It May Understate Risk​

Strengths of UBS’s approach:

• Grounded intelligence: Direct partner and customer conversations provide leading indicators of intent that supplement booking and revenue data.
• Actionable nuance: UBS identifies where demand is converting — reserved capacity, multi‑year contracts, AI GPU demand — rather than treating cloud spending as a monolithic metric.

Where UBS may understate or underweight risks:

• Conversion risk: Booked RPOs and long‑dated reservations can create optimism that is premature if accelerator supply, power, or construction bottlenecks delay deployments. The path from intent to recognized revenue remains the operational gating factor.
• Concentration and regulatory risk: Large perceived commitments (mega‑deals) sometimes make headlines before counterparties and contract terms are publicly disclosed; regulatory or geopolitical hurdles (data sovereignty, procurement restrictions) can stall public sector and cross‑border deals.
• Margin timing: Building AI capacity is capex‑heavy and can compress margins in the near term; higher‑margin AI services can only offset that pressure after scale and utilization improve. Investors should not assume margin expansion is automatic.

UBS’s messaging is measured, not sensational. That restraint matters: the firm signals durable demand but repeatedly cautions that the conversion path and capex execution are the watchpoints that will determine whether upbeat bookings translate into sustainable revenue and profits.

---

Bottom Line — Practical Takeaway for Readers​

• Corporate cloud spending looks stable and constructive, driven by AI and data modernization priorities, not immediate cuts. UBS’s field checks — direct customer and partner conversations at Oracle AI World — provide granular reasons to believe the hyperscalers will benefit.
• The winners will be those who can convert bookings into recognized revenue efficiently: Microsoft’s integration advantage, Google’s tooling and growth trajectory, and AWS’s scale all offer different risk/reward tradeoffs.
• Near‑term investor focus should be on capex cadence, RPO conversion rates, accelerator supply disclosures and gross‑margin impacts — the metrics that will validate UBS’s thesis in corporate filings and the upcoming earnings releases on October 29–30, 2025.

The UBS field checks add useful, ground‑level color to the market narrative: corporate IT spending is not surging wildly, but neither is it collapsing. Instead, budgets are becoming targeted — focused on AI and data workloads that favor cloud platforms. That’s a pragmatic environment for hyperscalers, provided they can execute on capacity, supply and monetization.

---

Conclusion
UBS’s on‑the‑ground conversations at Oracle AI World revealed a market that is cautiously optimistic and practically oriented: stable corporate cloud spending, a healthy appetite for AI accelerators, and distinct competitive vectors for Amazon, Google and Microsoft. Those dynamics support a reasoned allocation into the large hyperscalers while keeping a vigilant eye on execution signals — capex, accelerator supply, RPO conversion, and margin dynamics — that will determine whether “stable and healthy” evolves into consistent and visible revenue and profit acceleration across the cloud leaders.

---

Source: 富途牛牛 UBS Group: Corporate cloud spending remains stable and healthy, with Amazon (AMZN.US), Google (GOOGL.US), and Microsoft (MSFT.US) set to benefit.