I've gone from writing about SD-WAN to depending on it

No market-moving financial data: the article is a first-hand account of using SD-WAN to manage home internet instability across T-Mobile 5G and Starlink. The author reports roughly a dozen dropped connections over two months, but SD-WAN WAN failover prevented call/video disruption, and QoS/policy-based routing helped prioritize mission-critical traffic. Cost is cited at about $125/month versus prior fiber latency under 5ms (now ~15–80ms), while noting fiber expansion could eventually replace 5G/Starlink subscriptions.

Analysis

This reads as a small but meaningful validation of the "always-on edge" thesis: households and tiny offices in underwired geographies are increasingly assembling their own multi-link network stack, which shifts value from the access pipe to the routing/orchestration layer. That is constructive for TMUS because fixed wireless access monetizes excess network capacity with far less capex than fiber, but the real second-order winner is UI, where multi-WAN and policy-routing features create an attach opportunity for higher-end gateways and security gear.

The catch is that this is a reliability story, not a raw-speed story. If wireless and LEO links keep failing over to each other, the service that wins the primary relationship may still be the one that best supports redundancy and self-healing; that favors vendors with simple edge UX more than pure connectivity brands. HPE has a longer-cycle benefit only if SMB/branch SD-WAN refreshes reaccelerate, but that is budget-cycle dependent and unlikely to move the tape in the next quarter.

Contrarian view: the market may be overestimating the permanence of the wireless substitution. The moment fiber shows up, the primary links revert to backup status, which caps lifetime value and could pressure churn metrics if fiber buildouts accelerate faster than expected over 6-18 months. For TMUS, the key falsifier is rising FWA congestion/churn in the next 1-2 earnings prints; for UI, it is channel inventory normalization or weaker SMB spend rather than demand destruction from the concept itself.

Key Decisions for Investors

• Long TMUS on pullbacks into the next FWA subscriber update; 1-3 month thesis is low-capex broadband share gain, but exit if churn or congestion language worsens. Target is a modest re-rate, not a multi-bagger.
• Buy UI selectively on weakness rather than momentum; the setup favors a premium hardware/edge vendor, but the multiple already discounts some of the prosumer networking adoption. Falsify if next quarter shows softer channel demand or inventory build.
• Use HPE as a watch item, not a conviction long: only become constructive if Aruba/edge commentary confirms SD-WAN-driven refresh spending. Otherwise, the catalyst is too slow to matter over the next 1-2 quarters.
• If you want a relative-value expression, prefer long UI vs. a broad networking basket on any sector pullback; the thesis is better monetization of multi-WAN complexity, with downside limited by product breadth and upside tied to SMB adoption.