Selectel’s AI Direction Puts Storage Economics and API Access at the Center

November 17, 2026

Illustratie bij Selectel’s AI Direction Puts Storage Economics and API Access at the Center

AI demand meets a renewed role for high-capacity storage

Three Selectel articles published on 16 November 2026 connect AI adoption with changing infrastructure economics. Selectel reports, citing external market sources, that a 30 TB enterprise TLC SSD cost $22,600 in the third quarter of 2026, compared with $1,216 for an HDD of the same capacity; it attributes the widening pressure on storage prices largely to demand from AI data centers. The article argues that HDDs therefore remain important for high-volume storage, even though their own prices have risen, and highlights manufacturers’ plans for 44 TB, 50 TB and 55 TB drives. A second article describes Alibaba’s Qwen4-Max as a cloud- and API-only multimodal mixture-of-experts model with a one-million-token context window, aimed at coding, agentic work and project prototyping. Selectel presents benchmark and token-consumption results supplied by Alibaba while also noting that the model’s weights remain closed. The third article examines Xiaomi’s first two-bay NAS, including up to 60 TB of raw capacity, 2.5GbE networking and integration with HyperOS devices, but says that sales had only been announced for China.

A cloud strategy shaped by data gravity, model choice and constrained hardware

Taken together, the articles suggest that Selectel sees the next phase of AI infrastructure as more than a race for accelerators. Model access, economical data retention and movement between local devices, dedicated infrastructure and cloud APIs are becoming interdependent concerns. Qwen4-Max illustrates a shift toward powerful models consumed as metered services rather than deployed by customers, while the storage analysis indicates that the data supporting those services can make capacity planning materially more expensive. The NAS article adds an edge dimension: backup, media management and automated content search are moving into integrated local appliances, even if the featured product is consumer-oriented. Selectel’s references to its AI router, storage systems and dedicated servers align commercially with this interpretation. The emerging direction is a layered environment in which organizations select models through APIs, reserve flash for latency-sensitive workloads and retain larger data volumes on comparatively economical HDD-based tiers.

Benchmark, supply and regulatory assumptions need testing

The evidence has important limits. The storage prices are time-specific figures reproduced by Selectel from external research and company disclosures; they do not establish the prices, lead times or capacity commitments available from Selectel itself. Forecasts for component shortages and future drive generations are not guarantees of delivery or lower cost. Qwen4-Max performance figures include vendor-published and internal benchmarks, and the article identifies no direct comparison between the preview and final versions; benchmark leadership also varies by test. Closed weights create dependence on Alibaba Cloud’s API, pricing and service terms. Xiaomi’s NAS had no announced international release in the article, limiting its immediate relevance to Eastern European procurement. None of the selected articles sets out Selectel-specific service levels, data-location controls, model governance or migration arrangements. They also do not address sanctions, export controls, payment constraints or cross-border support, so organizations operating in or connected to Russia must validate legal eligibility and supply-chain feasibility separately rather than infer them from technical availability.

Decision guidance and sources: Selectel’s analysis of AI-driven HDD and SSD pricing

For Eastern European IT decision-makers, the practical conclusion is to treat AI architecture and storage procurement as one planning exercise. Build cost scenarios that separate hot flash, capacity HDD and backup tiers; test model quality and token consumption on representative workloads rather than relying on headline benchmarks; and preserve portability where a flagship model is available only through one provider’s API. Local or edge storage can support resilience and reduce unnecessary transfers, but the Xiaomi example should be viewed as a technology signal rather than a currently established regional procurement option. Buyers should request Selectel-specific evidence on availability, pricing, service levels, data handling and exit procedures, then conduct sanctions and supplier checks appropriate to their jurisdiction. The underlying sources are the storage-market assessment, Selectel’s Qwen4-Max release analysis and the Xiaomi Smart Storage overview.

Get the Cloud Insights weekly roundup

The week's most important stories on cloud, sovereignty and infrastructure, every Thursday in your inbox. Free.

We will email you to confirm first. Unsubscribe any time. Privacy policy

Leave a Comment