Dell Senior Director of Competitive Intelligence Jon Hyde has published three new AI Factory blog posts targeting key rivals including VAST Data.
The first post, “Where AI Factories Hit Their First Ceiling”, states Dell PowerScale delivers equivalent Nvidia AI performance while using 72% less power, 80% less rack space and 8x fewer backend switches than competing reference designs. Hyde explains Nvidia-powered AI Factories frequently run into data center power and space bottlenecks due to heavy GPU, storage and networking hardware demands.
He contends embedded AI storage platforms such as VAST Data’s AI OS require far more backend switches due to disaggregated scaling architecture. Additional switches raise rack occupancy, cabling, cooling and power costs before any GPU workloads even run.
Hyde stresses that enterprises should evaluate how much power budget their storage architecture consumes.
Per Dell’s testing aligned with Nvidia reference designs, PowerScale matches Everpure and VAST performance with far lighter infrastructure overhead. Against VAST specifically, it delivers identical performance with 41% lower power use and nearly 50% less rack space.
Hyde also challenges VAST’s public efficiency claims. He clarifies VAST’s touted 77% power savings and 73% space reduction from BlueField DPU offloads reflect internal upgrades over its older stack, not competitive advantages against industry-standard Nvidia reference architectures.
Hyde’s second blog, "It's Called a Database. It Doesn’t Act Like One,” argues that platforms strong in metadata and vector workloads cannot fully qualify as enterprise-grade databases. Structured business-critical table, record and time-series data requires robust capabilities that pure indexing-focused systems cannot satisfy.
He cites June 2025 theCUBE Research analysis describing VAST DataBase as essentially a “distributed index” lacking mature SQL optimization, cost-based query planning, role-based governance and comprehensive BI tool integration offered by enterprise platforms like Snowflake and BigQuery. The report concluded that despite fast revenue growth, VAST has not yet achieved lakehouse maturity comparable to Databricks or cloud database standards set by Snowflake and major hyperscalers — a view reaffirmed by theCUBE in February 2026.
Hyde further references NAND Research findings: while VAST supports open formats such as Apache Iceberg, its core database engine is proprietary and vertically customized. This tightly integrated design differs from the composable, open-ecosystem approach adopted by Dell, NetApp, HPE and Everpure, placing VAST closer to HCI-style closed architecture.
In contrast, Dell natively supports Apache Iceberg on ObjectScale, enabling seamless interoperability between PowerScale/ObjectScale-stored data and mainstream Databricks and Snowflake workloads without data migration.
Hyde’s third post, "The Blueprint for the Next Evaluation,” outlines five core evaluation criteria for IT buyers assessing AI data platforms, covering data gravity adaptation, operational overhead, GPU utilization, physical footprint and analytics stack compatibility.
Hyde states AI storage bidding evaluations should focus on real-world operational factors beyond benchmark demos: selecting architectures compatible with actual enterprise data gravity; fully accounting for sync jobs, staffing costs, GPU efficiency and hardware footprint; and preserving existing analytics toolchains rather than forcing workload migration.
He notes vendor sales decks target executive stakeholders, while facility, data engineering and FinOps teams bear actual operational costs — requiring cross-team participation in vendor evaluations.
Hyde also proposed five targeted questions to reveal vendors’ true architectural strengths: the three-year projected in-namespace data proportion; steady-state sync job and staffing requirements; reproducible open-model TTFT, token throughput and cache hit rate metrics; fully documented power, space and switch consumption at target scale; and native compatibility with existing Databricks and Snowflake deployments.
VAST Data declined detailed comment, stating only that it does not respond to competitor content.
Comment: Dell’s release of six detailed critique-focused blogs targeting VAST signals intense competitive pressure from the fast-growing rival data infrastructure vendor.
Beijing Qianxing Jietong Technology Co., Ltd.
Sandy Yang/Global Strategy Director
WhatsApp / WeChat: +86 13426366826
Email: yangyd@qianxingdata.com
Website: www.qianxingdata.com/www.storagesserver.com
Business Focus:
ICT Product Distribution/System Integration & Services/Infrastructure Solutions
With 20+ years of IT distribution experience, we partner with leading global brands to deliver reliable products and professional services.
“Using Technology to Build an Intelligent World”Your Trusted ICT Product Service Provider!
Your message must be between 20-3,000 characters!
English
