Vast Data Raises $1 Billion at $30 Billion Valuation as Storage Emerges as AI Infrastructure Bottleneck
Nvidia-backed data platform triples valuation in 16 months, underscoring investor pivot from GPUs to full-stack infrastructure as cluster scale demands efficiency over raw compute.
Vast Data has raised $1 billion at a $30 billion valuation, more than tripling its late 2023 valuation of $9.1 billion and marking one of the largest AI infrastructure rounds outside the chip sector. The Series F, led by Drive Capital and Access Industries with participation from Nvidia, signals a strategic shift in institutional capital toward storage and data movement infrastructure as large-scale compute clusters hit physical bottlenecks.
The raise comes as AI Infrastructure spending approaches $7 trillion through 2030, with storage and memory bottlenecks increasingly limiting GPU utilisation in hyperscale training clusters. Vast’s platform addresses data movement constraints that have emerged as a critical chokepoint: without efficient storage architecture, clusters with millions of GPUs cannot sustain the throughput required for frontier model training.
From GPU-Centric to Full-Stack Infrastructure
The valuation jump reflects a broader recalibration in AI capital allocation. While GPU manufacturers captured early infrastructure investment, operators of large-scale clusters now face diminishing returns from raw compute expansion. Data Center Dynamics reports that legacy storage architectures introduce latency that undermines GPU efficiency, particularly in distributed training workloads where data must move between thousands of nodes.
Vast’s DASE (Disaggregated Shared Everything) architecture collapses traditional storage silos, allowing compute and storage resources to scale independently while maintaining single-namespace simplicity. The company reports a Rule of X score of 228%—a composite of growth rate and profitability margin—indicating both revenue acceleration and operational efficiency, per the company. For comparison, most high-growth SaaS companies operate in the 80-120% range.
“We are already supporting AI environments spanning millions of GPUs globally, operating across every layer of the AI stack. What is becoming clear is that these layers are no longer independent. Applications, models, and infrastructure now operate as a single system through data.”
— Renen Hallak, Founder and CEO, Vast Data
Strategic Customer Base Spans Cloud, Enterprise, and Model Labs
Vast’s customer roster includes GPU cloud providers CoreWeave and Crusoe, AI model developers Mistral AI and xAI, enterprises like Lowe’s and JPMorgan Chase, and creative studios including Pixar. The U.S. Air Force also uses the platform for defence applications. Crusoe Cloud reported 10x partnership growth with Vast over the past year, underscoring adoption velocity among infrastructure providers building out AI-optimised data centres.
Nvidia’s participation in the round extends a strategic relationship that has deepened as the chip designer confronts storage as a constraint on cluster performance. Jensen Huang has publicly stated that data movement—not compute speed—now represents the primary bottleneck in large-scale AI systems, particularly as model parameter counts push into the trillions and training datasets exceed petabyte scale.
The round included more than $500 million in secondary sales, providing liquidity to employees and early investors, according to Ctech. Secondary shares traded at a 15% discount to the primary valuation, reflecting typical early-stage liquidity premiums. Vast employs approximately 1,100 people, with roughly one-third based in Israel.
Storage Efficiency Over Speed as Infrastructure Matures
The raise arrives as AI infrastructure investment pivots from capacity expansion to utilisation optimisation. Data Center Dynamics notes that power constraints, cooling limitations, and physical space are forcing operators to prioritise data reduction and movement efficiency over raw storage throughput. Vast’s architecture uses inline data reduction techniques that compress datasets without sacrificing retrieval speed—critical for iterative training workflows where the same data must be accessed repeatedly across distributed nodes.
- Vast Data’s $30 billion valuation represents a 3.3x increase in 16 months, driven by $4 billion in cumulative bookings and $500+ million in committed ARR.
- The raise underscores investor recognition that storage infrastructure—not just GPUs—is critical to scaling AI clusters beyond current bottlenecks.
- Customer base includes GPU cloud providers, AI model labs, and enterprises, reflecting broad adoption across the AI stack.
- More than half the $1 billion raise went to secondary sales, providing employee and early investor liquidity while signalling confidence in long-term value creation.
Chris Olsen, co-founder of lead investor Drive Capital, framed the valuation as recognition of Vast’s position at the centre of a “new class of infrastructure company” emerging from AI adoption. The firm cited both current momentum and structural market position as drivers of the step-change in valuation.
What to Watch
Vast’s ability to maintain its Rule of X performance as it scales will determine whether the $30 billion valuation proves sustainable or inflated. The company has not disclosed a path to IPO, but the secondary component of this round suggests preparation for eventual public markets. Near-term indicators include expansion of Nvidia co-engineering partnerships, adoption by additional hyperscalers (AWS, Microsoft, Google remain notable absences from the customer list), and penetration into inference workloads as deployment-side infrastructure spending accelerates. Competition from established storage vendors adapting legacy architectures for AI workloads—NetApp, Pure Storage, Dell—will test Vast’s architectural advantage as those players leverage installed enterprise bases. Watch for disclosure of gross margins in future rounds; high-growth infrastructure companies often sacrifice profitability for land-grab market share, and Vast’s Rule of X score suggests it may be threading that needle more successfully than peers.