Snowflake Surging to All-Time Highs

Snowflake shares rose more than 20% after revenue growth accelerated to 37% versus 35% expected, adjusted operating margin expanded to above 15% from 11%, and management forecast 36% revenue growth for fiscal 2027. AI workloads drove roughly half of the growth acceleration, while net revenue retention reached 126%, though the stock trades at roughly 20x forward sales and stock-based compensation diluted shares by more than 4% over the past year. Separately, opposition to data-center development is increasing: 833 U.S. groups reportedly blocked or delayed two-thirds of targeted projects in the first half, creating risks for AI-infrastructure suppliers despite expectations that state-level negotiations and hyperscaler funding could ultimately resume development.

Analysis

SNOW’s rerating now hinges less on another consumption beat than on whether AI-driven usage remains incremental rather than cannibalizing conventional analytics spend. The key transmission mechanism is operating leverage: if growth sustains while infrastructure costs and stock compensation decline as a share of revenue, consensus free-cash-flow estimates can compound faster than revenue. Conversely, a modest deceleration in consumption growth would be punished disproportionately at an elevated sales multiple; dilution remains a material drag on per-share value even if headline margins improve.

Data-center permitting friction is not a demand-destruction signal; it is a shift of value from construction throughput toward power-secured, politically viable capacity. Hyperscalers can absorb higher interconnection, generation, and community-benefit costs, but the resulting capex inflation pressures returns on incremental AI spend and favors vertically integrated operators. Near term, project delays create revenue-recognition and working-capital risk for MRVL, CLS and STRL; over 6-18 months, constrained supply can preserve pricing for suppliers with booked capacity, while lower-quality, single-project exposed contractors face order deferrals.

The consensus error is treating every infrastructure drawdown as an opportunity to buy beta. If the bottleneck migrates from chips to power, land and permits, the relevant question becomes which vendors have contractual visibility into energized sites—not aggregate announced capex. NVDA remains relatively insulated during the next 1-3 months because deployed capacity still requires accelerated compute, but a hyperscaler shift toward self-generation and longer construction schedules would push server shipments and networking recognition outward rather than eliminate it.

Key Decisions for Investors

• Maintain a tactical long SNOW only on post-earnings consolidation rather than chase momentum; target a 3-6 month hold. Require evidence that consumption growth remains above guided levels and net revenue retention stabilizes or rises; exit/reduce on a material guidance reset or renewed share-count growth above roughly 4% annually.
• Express the infrastructure bottleneck through a relative-value position: long NVDA / short a basket of MRVL and CLS, sized beta-neutral, for 1-3 months. NVDA has nearer-term demand visibility, while component and manufacturing names are more exposed to shipment timing, inventory digestion and delayed site energization; cover if hyperscaler capex guidance accelerates materially or MRVL/CLS order visibility improves.
• Avoid adding to STRL solely on a data-center recovery thesis until backlog disclosures distinguish awarded work from projects awaiting permits, grid interconnection or financing. A 6-18 month opportunity may emerge if power-ready project starts convert, but the missing data is site-level revenue timing and customer concentration.
• Monitor state-level power-cost allocation, interconnection rules and hyperscaler commitments to dedicated generation as the decisive catalyst. Binding cost-sharing agreements would be constructive for GOOG and META by de-risking build schedules, but also imply higher ongoing infrastructure cost and should cap multiple expansion if AI monetization does not accelerate in parallel.