RH
AI-native data platform combining all-flash storage, a unified namespace, and a built-in database to power exabyte-scale workloads at AI clouds, hyperscalers, research labs and enterprises.
A reasoned read from public sources. Each point links to its source.
VAST is one of the most credible AI infrastructure companies I've seen at this stage: the founding team built and scaled XtremIO to $1B+ in revenue before starting VAST, the architecture is technically differentiated (not just a rebrand), and the traction metrics — FCF positive, $2B ARR, xAI and CoreWeave as anchor customers — are genuinely exceptional. The risk is valuation and timing: at $30B, you're buying into a company that must become one of the defining infrastructure platforms of the AI era, not just a successful storage vendor. The secondary-heavy Series F and the breadth of the 'AI OS' ambition are the two things I'd probe hardest. But if you believe AI infrastructure spending is a decade-long secular trend and VAST can hold its architectural lead, this is a rare company worth serious consideration.12345
VAST operates at the intersection of AI infrastructure and enterprise data management — a market the company itself describes as a '$100T AI market.' Demand is being driven by hyperscale GPU clusters (xAI's 200,000-GPU Colossus, CoreWeave's $1.17B agreement) and the broader buildout of AI data centers. The Futurum analyst notes VAST's central role in 'supporting millions of GPUs,' and the company's $2B estimated ARR by end of 2025 suggests the market is real and already monetizable at scale. The tailwind is structural: AI training and inference pipelines require unified, high-throughput data infrastructure that legacy storage vendors were not built to provide.5673
VAST's core architectural innovation is DASE (Disaggregated and Shared Everything), which separates stateless compute nodes from shared NVMe flash enclosures, allowing any controller to access any data or metadata on any SSD — a meaningful departure from shared-nothing architectures used by most competitors. The platform has expanded from a storage product into a full 'AI Operating System' integrating DataStore, DataBase, DataSpace (global namespace), DataEngine (compute), and a Catalog layer for metadata. This breadth of integration — storage, database, and distributed compute in one platform — is the basis for VAST's 'N=1' positioning as the de facto AI data layer. The whitepaper-level technical depth suggests genuine engineering substance, not just marketing rebranding.8910115
CEO Renen Hallak previously led XtremIO's architecture from inception to over $1B in revenue as VP R&D, managing 200+ engineers — a directly relevant pedigree for building enterprise storage at scale. Co-founder Shachar Fienblit was CTO/VP Engineering at Kaminario and a storage architect at IBM for 8 years. Co-founder Alon Horev (CTO) brings database and storage platform experience from Cisco and IBM, plus Israeli Defense Forces software/hardware work. President Michael Wing came from Dell EMC. The founding team has unusually deep, directly applicable domain expertise across storage architecture, R&D leadership, and enterprise go-to-market.11213149
The traction evidence is exceptional for a private company: $4B+ in cumulative bookings, $500M+ in committed ARR (with industry estimates suggesting ~$2B actual ARR by end of 2025), positive operating margin, and positive free cash flow — the company states it does not require external capital to fund operations. Revenue reportedly tripled year-over-year. Landmark customer wins include xAI's 200,000-GPU Colossus cluster and a $1.17B agreement with CoreWeave. The Series F valuation tripled from $9.1B to $30B in roughly 2.5 years, with NVIDIA as a strategic investor. The company ranked #5 on Deloitte Technology Fast 500 in 2022 and #3 on Financial Times fast-growth list in 2023.46214375
VAST's named competitors include DDN (the incumbent in HPC/AI storage), Pure Storage (FlashArray/FlashBlade), Weka.IO (a direct parallel-file-system rival for AI workloads), and legacy vendors like Dell EMC, HPE, IBM, and Hitachi. In the PeerSpot rankings, VAST is #10 in NVMe All-Flash Arrays and #17 in File and Object Storage — respectable but not dominant in traditional storage mindshare. The more dangerous competitive threat is from hyperscalers (AWS, Google, Azure) building proprietary AI storage layers, and from NVIDIA itself potentially verticalizing storage into its AI stack — though NVIDIA's participation as an investor in VAST's Series F is a notable signal of partnership over competition for now. DDN and Weka.IO are the most direct pure-play AI storage rivals.1516171814
VAST has achieved something rare: hypergrowth (revenue tripling YoY) combined with operating profitability and positive free cash flow at scale, suggesting genuine product-market fit rather than growth bought with losses. The NVIDIA strategic investment and landmark wins with xAI and CoreWeave — the most demanding AI infrastructure operators on the planet — serve as powerful third-party validation of technical superiority. The DASE architecture's true shared-everything design creates a defensible moat: customers who build their AI pipelines on VAST's unified DataStore/DataBase/DataEngine stack face significant switching costs. At ~$2B ARR with a 15x revenue multiple, the $30B valuation is aggressive but not absurd for a category-defining infrastructure platform.62354
The $30B valuation prices in a very long runway of continued hypergrowth — any deceleration in AI infrastructure spending (macro shock, model efficiency gains reducing data demands, or hyperscaler capex pullback) would compress this multiple severely. More than $500M of the $1B Series F was secondary capital going to early investors and employees, not into the company — a sign of liquidity pressure among early backers that deserves scrutiny. The competitive moat in storage is historically difficult to sustain: Dell EMC, Pure Storage, and hyperscalers have deep distribution, balance sheets, and existing customer relationships. VAST's mindshare rankings (#10 in NVMe arrays, #17 in file/object storage) suggest it has not yet achieved dominant market share in its core categories. The 'AI OS' positioning also risks overreach — expanding from storage into database and compute puts VAST in direct competition with a much broader set of entrenched vendors.6141715
VAST needs to sustain revenue growth at or near its current trajectory to justify the $30B valuation — which means continuing to win the largest AI infrastructure deployments globally as xAI, CoreWeave, and their peers scale. The 'AI OS' platform expansion (DataBase, DataEngine, Catalog) must achieve genuine adoption beyond storage, creating the cross-sell and land-and-expand motion needed to grow ACV per customer. VAST must also successfully navigate the transition from serving a handful of hyperscale AI labs to broader enterprise adoption — a go-to-market motion that is structurally different and historically where infrastructure companies stall. Finally, NVIDIA's strategic alignment must deepen into a durable partnership rather than a hedge, as NVIDIA's own infrastructure ambitions could shift.56710
If you're a senior infrastructure or distributed systems engineer, VAST is one of the most compelling opportunities in the market right now. The technical problems are genuinely hard (global shared-everything storage at GPU-cluster scale), the team has a proven track record of building and shipping at this level, and the company is already profitable — so you're not betting on a speculative future. The catch is valuation: at $30B with a secondary-heavy last round, your equity upside requires VAST to become a $100B+ company, which means winning not just AI hyperscalers but broad enterprise. That's a real but non-trivial ask. Join if you want to work on foundational AI infrastructure with a team that has done it before; be clear-eyed that the easy valuation gains are behind you.21346
AI-native data platform combining all-flash storage, a unified namespace, and a built-in database to power exabyte-scale workloads at AI clouds, hyperscalers, research labs and enterprises.
Founded in 2016 by Renen Hallak (ex-XtremIO/Dell EMC), Shachar Fienblit (ex-Kaminario), Jeff Denworth (ex-CTERA) and Alon Horev. The team wanted to break the legacy storage tradeoff between performance and capacity, and built a disaggregated, all-flash architecture that has since become the backbone for the largest AI training and inference clusters.
Drive Capital and Access Industries co-lead the round with NVIDIA, Fidelity and NEA participating. Tripled the prior $9.1B valuation.
CNBC reports on VAST Data closing its Series F at a $30B valuation with NVIDIA as a participating investor.
VAST Data provides the storage backbone for xAI's Colossus supercomputer training Grok models, reducing TCO by 50%.
CoreWeave becomes the largest VAST customer, surpassing the xAI Colossus deployment.
Fidelity Management & Research Company leads the round, valuing the AI infrastructure company at over $9 billion.
Co-Founder & VP R&D
Previously VP R&D at Kaminario
Storage industry veteran and Kaminario alum who co-founded VAST Data alongside Renen Hallak.
Co-Founder & VP Technology
Previously Engineer at XtremIO
Co-founder and VP Technology at VAST Data.
$1.4B raised total
Petitioner on record
VAST DATA INC · NEW YORK, NY
FY 2025
See full record