Revision #1

VAST Data, the New York-headquartered storage infrastructure company founded in Israel in 2016, announced on April 22 the close of its Series F financing at a $30 billion valuation — more than three times the $9.1 billion it commanded in its December 2023 Series E round [1][2]. The round totals approximately $1 billion and was led by Drive Capital, with Access Industries as co-lead. Existing investors Fidelity Management & Research Company, New Enterprise Associates (NEA), and Nvidia also participated [1][3].

The deal ties the record set by Israeli cybersecurity firm Wiz for the largest single funding round by an Israeli-founded company [4]. But the headline number obscures a critical detail: more than $500 million of the $1 billion is secondary capital — money paid to early investors and employees selling their shares, not cash flowing into the company's treasury [5][6].

The Valuation Math

At $30 billion, VAST Data's valuation demands scrutiny against its actual financial performance.

The company reported reaching $200 million in annual recurring revenue (ARR) by January 2025 [7][8]. By the end of fiscal year 2025, VAST disclosed more than $500 million in Committed Annual Recurring Revenue (CARR) — a forward-looking metric that counts signed but not yet recognized contracts — along with positive operating margin and free cash flow [1]. The company projects $600 million in ARR for 2026 [5].

Source: Crunchbase, VAST Data Press Releases

Data as of Apr 22, 2026CSV

Against current revenue of roughly $500 million CARR, a $30 billion valuation implies approximately a 60x multiple on committed revenue, or roughly 15x if measured against the more optimistic near-term ARR trajectory the company is projecting. Either way, the premium is steep relative to publicly traded storage companies.

Source: Yahoo Finance, Company Filings

Data as of Apr 22, 2026CSV

Pure Storage (now branded Everpure) trades at approximately 6.7x trailing revenue on a market capitalization of $22–24 billion, with $3.5–3.7 billion in trailing twelve-month revenue [9]. NetApp commands a roughly 3.1x multiple on its $6.7 billion in revenue and an $20.6 billion market cap [9]. VAST Data's implied multiple is at least double Pure Storage's and roughly five times NetApp's.

The gap is not inherently unreasonable for a company growing at VAST's pace — but it prices in years of sustained hypergrowth that have yet to materialize at public-company scale.

Where the Money Goes

The structure of the round itself tells a story. At $1 billion total, the financing buys roughly 3.3% of the company at the $30 billion valuation. But because more than half is secondary, the actual primary capital injection — new money for operations and growth — is less than $500 million [5][6].

VAST's press release states the secondary component "provides liquidity for early investors and employees" and supports "planned geographic expansion and ecosystem development" [1]. In practice, secondary-heavy rounds at this stage often signal that the company doesn't urgently need cash (VAST claims five consecutive years of positive free cash flow) but that long-tenured shareholders need an exit valve [6][10].

CEO Renen Hallak has indicated the company is preparing for an IPO, potentially by the end of 2026 or into 2027 [10][11]. CFO Amy Shapero, who joined from Shopify, estimated roughly two years to reach IPO readiness [11]. The secondary liquidity reduces pressure on the IPO timeline by letting early backers and employees take partial returns now.

The CoreWeave Concentration Question

VAST Data's most significant commercial relationship is with CoreWeave, the GPU cloud provider valued at approximately $35 billion. In November 2025, the two companies signed a $1.17 billion commercial agreement making VAST the primary data platform underneath CoreWeave's compute infrastructure [7][8][12].

The contract, which VAST said spans three to five years, covers software licenses for VAST's storage layer plus higher-level services including checkpointing, KV caching, streaming, and database functions [8]. CoreWeave's customers include Meta, OpenAI, and Microsoft, meaning VAST is effectively the storage backbone for some of the largest AI training operations in the world [8].

Source: Calcalist, Blocks and Files, Company Disclosures

Data as of Apr 22, 2026CSV

This deal alone likely accounts for a substantial share of VAST's revenue trajectory. The company has disclosed more than $4 billion in cumulative bookings [1], with the CoreWeave agreement representing roughly 29% of that total. VAST also serves as the data foundation for more than 30 AI cloud service providers worldwide [13].

Enterprise customers beyond the AI cloud segment include Lowe's, CVS Health, the U.S. Air Force, NASA, the U.S. Department of Energy, and the NHL [13]. But the degree to which non-AI enterprise revenue has scaled relative to AI-cloud revenue remains unclear from public disclosures.

The concentration risk is twofold. First, CoreWeave itself is a heavily leveraged company that has raised billions in debt to finance GPU purchases. If CoreWeave's own business contracts, VAST's largest revenue stream contracts with it. Second, if AI infrastructure spending broadly declines — whether due to efficiency breakthroughs, a macroeconomic downturn, or a pullback in venture funding for AI startups — the cohort of GPU cloud providers that make up VAST's customer base could see simultaneous demand reductions.

The DeepSeek Factor and AI Spending Durability

In January 2025, Chinese AI lab DeepSeek demonstrated that competitive large language models could be trained for a fraction of the cost previously assumed — reportedly $6 million using 2,000 Nvidia H800 GPUs, compared to the $80–100 million and 16,000+ GPUs estimated for models like GPT-4 [14][15].

The announcement briefly rattled AI infrastructure stocks and raised the question of whether the multi-hundred-billion-dollar AI buildout was oversized. But the subsequent market response has been more nuanced. Bloomberg Intelligence projected that hyperscale companies including Microsoft, Amazon, and Meta would spend $371 billion on data centers and AI computing in 2025 — a 44% increase from the prior year — with spending projected to reach $525 billion by 2032 [16]. Nearly 90% of data center executives surveyed said they did not expect to reduce capital expenditures in response to DeepSeek [17].

The bull case for VAST Data rests on the argument that efficiency gains in model training actually increase total demand for AI infrastructure, because lower per-unit costs make more use cases economically viable, generating more data and more inference workloads that require storage. The bear case holds that a subset of AI cloud providers — particularly the smaller, venture-funded "neoclouds" that make up a meaningful portion of VAST's customer base — are running on borrowed time and borrowed money.

S&P Global modeled a scenario in which AI inference infrastructure spending drops 30–50%, with per-provider capital expenditures falling from $80–100 billion to $65–85 billion annually [14]. In such a scenario, storage infrastructure providers like VAST would feel downstream effects, though the magnitude depends on contract structure and how much revenue is already committed versus pipeline.

Technical Architecture: What VAST Actually Sells

VAST Data's core technology is its Disaggregated Shared Everything (DASE) architecture, designed to eliminate what the company describes as a fundamental tradeoff in traditional distributed storage: the choice between scale, performance, simplicity, and cost [18][19].

In a DASE system, stateless compute nodes (called CNodes) connect to all-flash data storage nodes (DNodes) over high-speed NVMe-oF fabric (a protocol for accessing flash storage over a network). Every compute node can access the entire global dataset directly, eliminating the "east-west traffic" — internal node-to-node communication for locking, cache management, and metadata coherency — that limits scaling in legacy architectures [18].

The practical result, according to VAST, is a storage system that scales to exabytes of capacity and delivers terabytes per second of throughput at a cost that makes traditional storage tiering (moving data between fast and slow storage based on access patterns) unnecessary [18].

VAST has published internal benchmarks claiming single-millisecond latency, 20x faster queries and 40x faster transactions than legacy systems, and 100x acceleration on single-key database queries [19]. However, independent third-party benchmarks validating these claims at production scale are limited. VAST has achieved NVIDIA's Certified Storage designation, and Supermicro has launched an integrated solution with VAST, but these are partnership certifications rather than independent performance audits [13].

Incumbent competitors Dell EMC, NetApp, and Pure Storage each have their own all-flash and AI-optimized offerings. Dell holds over a quarter of the broader storage market and competes aggressively on price [20]. Pure Storage's FlashBlade product line targets similar AI and high-performance workloads [20]. The competitive moat for VAST depends on whether its architectural approach — built from scratch for AI workloads rather than adapted from legacy designs — delivers sufficient performance and cost advantages at scale to justify switching from established vendors.

Headcount, Cash Flow, and the Path Forward

VAST Data employs approximately 1,100 people, roughly one-third based in Israel [5][6]. The company has been consistently profitable on a free cash flow basis, claiming to generate more than $100 million in cash per quarter [10]. In its most recent fiscal year, VAST reported a "Rule of X" score — a metric combining revenue growth rate and profit margin — of 228% [1].

These are exceptional numbers for an enterprise infrastructure company at VAST's stage. If accurate, they suggest VAST is not a cash-burning startup subsidizing growth with venture capital, but a profitable business with strong unit economics. The $1 billion raise, in this framing, is less about survival capital and more about strategic positioning: bringing in Nvidia and other investors as ecosystem partners, providing employee liquidity, and building a war chest for potential acquisitions.

Hallak has stated that VAST "can grow faster as a private company with fewer distractions," but acknowledged that as the customer base shifts toward large enterprises, the dynamics may change [10]. The appointment of Shapero as CFO in 2024 — a former Shopify executive with public-company experience — signals that the IPO path is being actively prepared [11].

What $30 Billion Needs to Be True

For a public-market investor to achieve a reasonable return buying VAST Data at its current $30 billion valuation, the company would need to reach roughly $3–4 billion in annual revenue within 3–5 years while maintaining current profit margins. That would bring the forward revenue multiple into the range of what the public markets currently pay for high-growth storage companies like Pure Storage (6–7x).

VAST's disclosed growth rate — from $200 million ARR in January 2025 to $500 million+ CARR by fiscal year-end — suggests the company is roughly doubling annually [1][7]. Sustaining that pace would put VAST at approximately $2 billion in revenue by 2028 and $4 billion by 2029. That is plausible if AI infrastructure spending continues its current trajectory and VAST maintains its share of the AI storage market. It becomes much harder if AI spending plateaus or if incumbents like Dell, NetApp, and Pure Storage successfully compete for AI workloads with their own flash-optimized products.

The embedded term-sheet protections in the Series F round are not publicly disclosed. Late-stage rounds at these valuations typically include liquidation preferences — contractual provisions ensuring that Series F investors receive their money back (often with a guaranteed return) before earlier shareholders in a downside scenario such as an IPO below $30 billion or an acquisition. Whether VAST's round includes participating preferred shares, anti-dilution ratchets, or other protective provisions would significantly affect the risk-reward calculus for earlier investors.

The Broader Picture

VAST Data's $30 billion valuation is a bet that AI infrastructure is a generational buildout, not a cyclical bubble, and that the storage layer — traditionally a low-margin, commodity business — will be re-priced as a high-value platform in an AI-driven computing stack. The company's financial performance to date supports the growth thesis: positive cash flow, tripling revenues, and a $1.17 billion anchor contract with CoreWeave are tangible markers.

The risks are equally concrete: heavy customer concentration in a cohort of GPU cloud providers whose own financial stability is untested, a valuation premium that assumes years of sustained execution, and a competitive landscape where the largest technology companies in the world (Dell, with over 25% storage market share) are aggressively building their own AI storage capabilities [20].

Whether $30 billion proves prescient or inflated will depend less on VAST Data's technology — which appears genuinely differentiated — and more on whether the AI infrastructure spending cycle that produced it proves durable across economic conditions that have yet to be stress-tested.