Nutanix at Goldman Sachs conference: cloud push gains pace

Nutanix guided FY2027 operating margins to 24%-25% and reiterated a longer-term path above 30%, supported by strong bookings momentum and expansion beyond hyperconverged infrastructure into external storage, public cloud, Kubernetes, databases and AI. External storage—addressing the 80% of the installed base still using third-party storage—is expected to be the company’s largest FY2027 growth driver after multiple seven-figure Q4 FY2026 wins. The outlook remains tempered by server prices running 4-5x above year-ago levels, uneven hardware availability, delayed public-sector projects and a net bookings headwind from future start dates and slower renewal-pool growth. Nutanix also committed $20M to internal GPU clusters, expected to serve roughly 80% of its AI needs and pay back within about one year.

Analysis

NTNX’s upside case is less about AI monetization than reducing migration friction: external-storage compatibility turns a disruptive infrastructure replacement into a software overlay, potentially improving win rates and shortening sales cycles against Broadcom/VMware. This also weakens the traditional HCI hardware-refresh dependency, but it creates a more direct coopetition with NTAP and server OEM ecosystems. The near-term market risk is that investors capitalize the expanded TAM before evidence that external-storage bookings are recurring, attachable, and not merely pull-forwarded migration deals.

Over the next 1-3 months, deferred license commencements can create a gap between reported bookings and recognized revenue/ARR, making an already extended stock vulnerable to even unchanged guidance at the next print. Margin delivery is the key offset: incremental specialist sales capacity and cloud-native products should initially dilute operating leverage before attachment becomes material. The thesis is falsified if FY27 operating-margin guidance is reduced, future-start-date mix rises faster than expected, or external-storage deal sizes fail to translate into broad customer count growth.

The non-obvious beneficiary of enterprise AI cost optimization is AMD: heterogeneous, open-weight deployments improve the economic case for non-NVIDIA GPU capacity, particularly where customers prioritize inference cost and sovereign deployments over a proprietary CUDA stack. Conversely, NTNX’s AI gateway layer is likely to be competitively crowded and should not be assigned a stand-alone platform multiple until management discloses pipeline conversion, paid deployments, and retention metrics. Consensus may underappreciate the multi-year VMware displacement runway, but may also underappreciate that procurement friction can delay—not eliminate—software demand, producing lumpy quarterly results.

Key Decisions for Investors

• Maintain NTNX as a 6-18 month structural long only on a 10-15% pullback or after the next earnings release confirms external-storage growth and FY27 margin guidance; target a 20-30% upside from entry versus 12-15% downside if guidance is merely maintained. Do not chase a six-month momentum move without booking-to-revenue visibility.
• Use a pair trade: long NTNX / short AVGO over 6-12 months in equal dollar beta-adjusted sizing. The trade expresses incremental VMware workload migration, while AVGO’s downside is limited by its broader semiconductor mix; exit if Broadcom introduces materially more flexible VMware pricing or NTNX’s migration pipeline weakens.
• Watch NTAP rather than short it: external-storage interoperability can expand Nutanix’s addressable footprint while preserving installed storage arrays, making NTAP a potential second-order beneficiary. Upgrade to a tactical long only if NTAP reports improving enterprise array demand or explicitly identifies Nutanix-enabled wins.
• Add AMD on evidence that Nutanix’s enterprise AI partnerships generate deployed inference clusters, not announcements; a 6-12 month long is preferable to NVDA for this specific thesis. Falsifier: enterprise customers continue selecting NVIDIA-only stacks or AMD data-center GPU revenue/guidance fails to accelerate.