June 10, 2026 · 13 min read
Cloud computing is not a theme — it's the operating substrate of the global economy. The migration from on-premise to cloud is roughly 30% complete globally, and every AI workload accelerates it. Here's a full-metrics guide to the best plays across the stack.
Cloud computing shifts enterprise IT from CapEx (buy servers, own data centers, hire infrastructure teams) to OpEx (rent compute by the hour, scale instantly, pay only for what you use). The financial benefit is dramatic: capital previously tied up in data centers becomes free cash flow.
Three structural forces keep cloud growth at 20%+ for the foreseeable future: (1) Digital transformation — every industry is replacing legacy on-premise software with cloud-native SaaS applications. The global enterprise software market is 70% still on-premise. (2) AI requires cloud compute — training GPT-5 class models requires 100,000+ GPU clusters that only hyperscalers can build and operate. Every AI workload is a cloud workload. (3) Multi-cloud architecture — enterprises are running multiple cloud providers simultaneously (AWS + Azure + GCP) for redundancy and cost optimization, growing all three hyperscalers simultaneously.
The IaaS → PaaS → SaaS stack layers also create compounding revenue: once a company is on AWS EC2 (IaaS), they adopt S3 (storage), RDS (database), Lambda (compute), SageMaker (AI) — expanding spend by 2–5× over 5 years with minimal sales effort.
| AWS (AMZN) | Azure (MSFT) | Google Cloud (GOOGL) | |
|---|---|---|---|
| Market Share | ~33% | ~25% | ~12% |
| Revenue (annualised) | $105B+ | $130B+ | $50B+ |
| Revenue Growth | +17% | +29% | +28% |
| Operating Margin | 38% | 43% | 30% |
| AI Strategy | Bedrock + Trainium chips | Azure OpenAI + Copilot | Gemini + TPU silicon |
| Fastest-growing service | Bedrock (AI inference) | Azure OpenAI Service | Vertex AI + Firebase |
| Pricing vs AWS | Baseline | +5–10% premium | 15–30% cheaper |
AI scores use BriMindInvest's composite signal (20–96 scale). Op. margin shown for cloud/tech segment where reported separately. Data June 2026.
| Ticker | Tier | AI Score | Fwd P/E | Rev Growth | Gross Margin | Op Margin | Buy% | Target ↑ |
|---|---|---|---|---|---|---|---|---|
| AMZN | Hyperscaler | 88 | 38x | +17% | 49% | 38% | 92% | +15% |
| MSFT | Hyperscaler | 85 | 34x | +16% | 70% | 43% | 90% | +12% |
| GOOGL | Hyperscaler | 83 | 20x | +14% | 58% | 30% | 85% | +18% |
| DDOG | Cloud Observability | 76 | 75x | +25% | 80% | 22% | 78% | +18% |
| SNOW | Cloud Data Platform | 72 | 160x | +26% | 67% | — | 65% | +22% |
| CRM | SaaS | 74 | 26x | +9% | 78% | 30% | 76% | +20% |
Every AI model requires cloud compute to train, deploy, and scale. The hyperscalers are spending $80B+ each in 2026 on AI GPU clusters — the equivalent of building 10 new data centers per month. This capital expenditure creates a durable moat: the physical infrastructure for AI is being locked up for 10–20 years in hyperscaler-owned facilities.
The 10-year thesis: By 2035, the majority of enterprise software will be AI-augmented or AI-native. Every application will require inference compute — running through one of the three hyperscaler AI platforms. This is the most durable structural growth story in technology.
| Ticker | Name | ARR Growth | Gross Margin | NRR | Fwd EV/Rev | AI Product |
|---|---|---|---|---|---|---|
| SNOW | Snowflake | +26% | 67% | 127% | 15x | Cortex AI, Arctic LLM, data for RAG |
| DDOG | Datadog | +25% | 80% | 122% | 18x | LLM Observability, AI cost management |
| NET | Cloudflare | +27% | 79% | 115% | 20x | Workers AI, AI Gateway, inference routing |
| NOW | ServiceNow | +23% | 80% | 120% | 16x | AI workflow agents (Now Assist) |
| CRM | Salesforce | +9% | 78% | 104% | 7x | Agentforce — autonomous AI CRM agents |
| WDAY | Workday | +15% | 76% | 110% | 8x | Workday AI — HR + finance automation |
NRR measures how much existing customers spend year over year — inclusive of churn, contraction, and expansion. An NRR of 120% means that even if you never signed a single new customer, your revenue would grow 20% annually from existing accounts alone. This is the most powerful metric in SaaS because it directly measures product stickiness and upsell potential.
| Company | NRR | What it means |
|---|---|---|
| Snowflake (peak) | 170% | Existing customers spending 70% more per year — unprecedented |
| Snowflake (current) | 127% | Still strong: existing customers spending 27% more per year |
| Datadog | 122% | Module expansion driving upsell — 28+ product modules |
| Salesforce | ~104% | Mature SaaS: modest upsell; Agentforce AI agents the next lever |
| AWS | N/A | Consumption-based — customers spend more as they scale workloads |
NRR above 120% is rare and exceptional. SNOW's 170% peak NRR in 2022 was driven by enterprises spending 70% more every year — largely driven by data growth and Snowflake's consumption model. As cloud spending normalized, NRR compressed to 127%. Still strong, but showing the consumption model's revenue volatility vs. subscription SaaS.
The 2021–2022 era punished unprofitable cloud companies hard. The survivors that reached profitability while maintaining growth are the most attractive risk-adjusted investments in 2026:
Every company moving to cloud needs cloud-native security. Traditional firewall-based security was designed for on-premise perimeters — it doesn't work when data is spread across AWS, Azure, SaaS apps, and remote workers. Cloud-native security is replacing it:
Platformization trend: Both CRWD (Falcon) and PANW are attempting to replace point products with platforms — bundling 15–25 security functions in one platform. Enterprises want fewer vendors (vendor fatigue), making platform consolidation the dominant trend in enterprise security through 2028.
Azure (43%) best-in-class. AWS (38%) improving. Google Cloud (30%) expanding fastest. SNOW (—) still investing. Higher margins = more durable earnings in a downturn.
The three hyperscalers collectively spent $235B+ on AI-related capex in 2025 and are on track to spend $300B+ in 2026. Each runs a fundamentally different AI strategy:
AWS strategy: Vertical integration — Trainium chips for AI training, Inferentia for inference, Bedrock for model APIs. Broadest enterprise customer base; most AWS AI revenue comes from existing customers expanding workloads rather than new logos.
Azure strategy: OpenAI exclusivity. Azure OpenAI Service carries ~29% of all enterprise-deployed GPT-4/GPT-5 workloads globally. Copilot integration across 500M+ Microsoft 365 seats creates monetization leverage that no other cloud platform can match.
Google Cloud strategy: Gemini-native. GCP's differentiation is its Tensor Processing Units (TPUs) — the only hyperscaler with proprietary AI silicon competitive with NVIDIA at scale. Google Cloud's AI pricing is 15-30% below Azure for comparable inference workloads, driving share gains in price-sensitive enterprise segments.
Datadog (DDOG): The monitoring layer for cloud-native infrastructure. Its LLM Observability product captures AI model latency, cost, and error data — a high-demand product as enterprises struggle to manage AI inference costs. NRR consistently above 120%; 28 modules (vs. 16 three years ago) create continuous upsell surface within the existing base.
Snowflake (SNOW): The structured data layer for AI. Enterprises store their proprietary datasets in Snowflake to fine-tune models, run RAG pipelines, and feed AI applications. New CEO Ramaswamy is accelerating AI product velocity; the consumption-based model is volatile quarter-to-quarter but structurally sound. Highest risk/reward in this cohort.
Salesforce (CRM): Best value in this group at 26x forward earnings. Agentforce (AI agents for CRM automation) is generating strong early enterprise pipeline. At 30% FCF margin with a $5B+ annual buyback, Salesforce offers a rare combination of cheap multiple, AI option value, and shareholder return discipline.
| ETF | Name | Expense Ratio | Focus | 5yr Return |
|---|---|---|---|---|
| SKYY | First Trust Cloud Computing ETF | 0.60% | Cloud pure-plays + hyperscalers | +12.1% ann. |
| WCLD | WisdomTree Cloud Computing ETF | 0.45% | Pure-play SaaS/cloud software | +8.4% ann. |
| IGV | iShares Expanded Tech-Software ETF | 0.41% | Broad software (cloud + enterprise) | +15.2% ann. |
IGV has the best 5-year track record and reasonable 0.41% ER. WCLD provides the purest cloud software exposure but with higher volatility. SKYY includes both hyperscalers and SaaS — most diversified of the three. None matches the cost efficiency of broad market ETFs (0.03–0.04%) — active stock picking in cloud has historically justified the extra effort vs. these thematic ETFs.
For 2026, the best risk-adjusted cloud investments are: GOOGL (Google Cloud growing 28% while GOOGL trades at 20x — the cheapest hyperscaler by a wide margin), AMZN (AWS profitability + retail recovery + Bedrock AI upsell), and CRM (best value in SaaS at 26x with Agentforce AI catalyst and $5B+ FCF).
For pure AI cloud exposure, MSFT remains the premier OpenAI exclusivity play. For higher-risk/reward: SNOW under new CEO Ramaswamy is the turnaround story — the data platform for AI fine-tuning has structural advantages if execution improves. For passive exposure, IGV at 0.41% ER captures the full cloud software universe with reasonable cost.
Free side-by-side AI scores, operating margins, revenue growth, and analyst targets for any two cloud stocks.
Get AI prediction signals, unlimited stock comparisons, portfolio analytics, and personalized watchlists — free for 14 days, no credit card required.
14-day free trial · No credit card required · Cancel anytime