Vast Data closed a $1 billion Series F financing round at a $30 billion post-money valuation on April 22, 2026, a deal Reuters confirmed the same day, more than tripling the New York-based company’s $9.1 billion Series E mark from late 2023 – a jump from roughly $9 billion to $30 billion that CRN, in an April 2026 interview with the company, described as Vast’s largest funding round to date, and that Axios independently corroborated in its own coverage of the deal – crowning it the most valuable private data infrastructure software company in history. The round, co-led by Drive Capital and Access Industries with participation from NVIDIA, Fidelity Management & Research Company, and New Enterprise Associates, pushed Vast’s cumulative lifetime funding to roughly $1.38 billion, a figure Blocks & Files first disclosed in March 2026 ahead of the formal announcement, and included over $500 million in secondary capital, signaling that long-time employees and early investors are finally getting a path to liquidity ahead of an expected initial public offering.
The financing arrives in the middle of an AI infrastructure spending cycle that few precedents prepare investors for. Vast Data has now booked more than $4 billion in cumulative sales, holds over $500 million in committed annual recurring revenue at the close of fiscal year 2025, and is growing revenue at roughly triple the pace of last year while remaining cash-flow positive. The Series F documents value the platform powering xAI’s 200,000-GPU Colossus cluster and CoreWeave’s $1.17 billion deployment at a multiple that no traditional storage vendor has ever commanded.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Inside the $1 Billion Series F: Drive Capital and Access Industries Co-Lead a Tripled Valuation
Vast Data confirmed the financing in a press release issued from its New York headquarters on April 22, 2026, ending months of speculation about a round that had been in the works since the fourth quarter of 2025. Drive Capital, the Columbus-based growth fund founded by ex-Sequoia partners Mark Kvamme and Chris Olsen, and Access Industries, the privately held conglomerate of billionaire Len Blavatnik, anchored the deal as co-leads. NVIDIA, which has been on Vast’s cap table since the Series E and supplies the GPUs that Vast’s customers cluster around, increased its position. Fidelity Management & Research Company and New Enterprise Associates, both Series E participants, also wrote follow-on checks.
The structure of the round is the part that veteran venture watchers will dissect first. Of the roughly $1 billion gross, more than $500 million was secondary, meaning shares were sold by existing holders rather than newly issued by the company – a figure Calcalist independently confirmed in its March 2026 reporting – though earlier reporting from Globes in February 2026, relayed via Blocks & Files, had pegged the secondary component as high as $700 million to $850 million before the structure was finalized. Vast disclosed in supporting materials, which Blocks & Files reported in March 2026 as raising new Series F preferred capital at the $30 billion valuation, that the secondary tender was sized to give early employees and angel investors meaningful liquidity without forcing the company to issue heavy dilution at a richer valuation, a rationale CTech reported was aimed squarely at rewarding long-tenured staff and early backers. Primary proceeds will fund what chief executive Renen Hallak described as the company’s pivot from a storage vendor to an “AI operating system,” with new hires planned for product, MENA expansion, and federal go-to-market.
The valuation jump from $9.1 billion in December 2023 to $30 billion in April 2026 represents a 230 percent increase across roughly 28 months. Few private companies have re-rated that quickly in the AI era. Cursor’s Anysphere, which raced from a $400 million valuation in early 2024 to a reported $60 billion mark in 2026, is one of the only comparable trajectories. The difference is that Vast Data has actual recurring revenue that scales with GPU consumption, while many AI darlings still depend on a single product line tied to consumer subscriptions.
Why a Storage Company Is Worth $30 Billion in the AI Era
The simple answer is that Vast Data is no longer a storage company. The 2016-founded vendor sells what it calls a Disaggregated Shared Everything (DASE) architecture that physically separates compute nodes from storage media using NVMe over fabric. That choice, which sounded like a marketing exercise when the platform first shipped in 2019, turned out to be the exact shape that GPU clusters need. When 200,000 H100 or B200 GPUs simultaneously read training tokens or write checkpoint files, traditional dual-controller arrays from NetApp or Pure Storage saturate. Vast’s stateless compute layer scales horizontally to absorb the traffic without forcing customers to shard data across appliances.
That technical match-up converts into pricing power. Vast charges per usable petabyte under a perpetual license, then sells subscription support, software updates, and increasingly an AI workflow layer called the Vast AI Operating System. CoreWeave’s previously disclosed $1.17 billion contract, which spans multiple years and follows the cloud’s rapid GPU buildout, is the sort of deal that simply does not exist in the legacy storage market. NetApp’s largest publicly disclosed contract is in the high tens of millions; Vast routinely signs nine-figure agreements with neoclouds, sovereign AI projects, and government labs.
The financial profile justifies the multiple in a way most software companies cannot match. Vast disclosed a Rule of 40 score of approximately 228 percent in supporting materials, calculated as growth rate plus operating margin. The Rule of 40 is a long-standing SaaS heuristic in which a score above 40 percent is considered healthy. Snowflake hit roughly 70 percent at the peak of the cloud migration. Cohesity targeted the mid-30s prior to its Veritas acquisition. A Vast at 228 percent puts the company in a category by itself among private infrastructure vendors, and it is the single number that most investors point to when asked why they paid $30 billion for a storage company.
The Founders: Renen Hallak, Jeff Denworth, Shachar Fienblit, and Alon Horev
Renen Hallak founded Vast Data in 2016 alongside Jeff Denworth, Shachar Fienblit, and Alon Horev. Hallak previously led the development of XtremIO, the all-flash array that EMC acquired for $430 million in 2012 and turned into a multi-billion-dollar product line. Denworth, who serves as chief marketing officer, came from CTERA Networks; Fienblit, the chief research and development officer, also worked at XtremIO and Kaminario; and Horev, the chief technology officer, has a deep background in distributed storage from his Israel Defense Forces days. The four-founder team remains intact almost a decade later, an unusual achievement for any startup, let alone one that has crossed $4 billion in cumulative bookings – and Calcalist, whose English-language CTech desk carried the story internationally, called the $1 billion Series F the largest private equity round in Israeli high-tech history in its March 2026 coverage, a distinction that traces back to Hallak’s XtremIO pedigree and Horev’s roots in Israel’s IDF-trained storage engineering community.
“Our customers tend to buy big, and they tend to buy often,” Denworth said in remarks accompanying the funding announcement. “We’ve been deliberate about focusing on the upper end of the AI market, where the problems are more complex but also more consistent. That lets us grow quickly without fragmenting the product or the organisation.” The strategy reads as a deliberate rejection of the mid-market push that defined storage incumbents like Pure Storage and NetApp during the last decade. Vast’s go-to-market team is structured around a small number of very large accounts rather than a broad channel.
Hallak, for his part, played down the urgency of the raise in interviews following the announcement. “VAST doesn’t need the cash,” he told Fierce Network, pointing to the company’s positive free cash flow and tripled revenue. The framing is consistent with a board that wanted to clean up the cap table, give employees liquidity, and lock in a benchmark valuation ahead of an IPO that the company has telegraphed could land in late 2026 or 2027.
Vast Data Series F vs Prior Rounds: A Decade of Compounding
| Round | Date | Amount | Post-Money Valuation | Lead Investor |
|---|---|---|---|---|
| Series A | February 2019 | $40 million | Not disclosed | Norwest Venture Partners |
| Series B | April 2020 | $100 million | $1.2 billion | Next47 |
| Series C | May 2021 | $83 million | $3.7 billion | Tiger Global |
| Series D | December 2022 | $118 million | $9.1 billion | Fidelity |
| Series E | December 2023 | $118 million extension | $9.1 billion | NEA, NVIDIA |
| Series F | April 22, 2026 | $1 billion (incl. $500M+ secondary) | $30 billion | Drive Capital, Access Industries |
The capital trajectory tells the same story the income statement does. Vast raised modest sums during the pre-AI era, then broke out alongside the generative AI training boom that began in late 2022. The Series E was effectively a flat extension at $9.1 billion, a sign that even Vast was uncertain how durable the spending wave would be. By the end of 2025, with CoreWeave’s deployment ramping and xAI’s Colossus build pulling tens of millions of dollars per quarter, the company had the metrics to demand a tripled mark – a valuation leap from roughly $9 billion to $30 billion that both SiliconANGLE and Built In confirmed in their respective April 2026 coverage of the Drive Capital- and Access Industries-led round, with Built In likewise pegging the jump from $9.1 billion to $30 billion as one of the fastest re-ratings of the current AI infrastructure cycle.
The Customer Roster: CoreWeave, xAI, Lowe’s, US Air Force, and Cursor
Vast disclosed an unusually specific customer list to support the Series F valuation. CoreWeave, the publicly traded AI cloud that signed a $1.17 billion contract with Vast in 2024, remains the largest single account. Elon Musk’s xAI runs Vast Data inside the 200,000-GPU Colossus supercluster in Memphis, Tennessee, where the platform handles training token storage and checkpoint management for the Grok model family. Lowe’s, the home improvement retailer, uses Vast as the data layer for its supply chain and store analytics workloads. The United States Air Force adopted Vast for several classified programs through a Cloud One partnership. Cursor, the AI coding company that itself recently hit a reported $60 billion valuation, runs its codebase indexing and retrieval workloads on Vast.
That mix of neoclouds, sovereign AI labs, federal agencies, and enterprise AI adopters is what Vast points to when defending the valuation against bears who argue the platform is overly exposed to a small number of large GPU clouds. CoreWeave, while the biggest customer, is reportedly less than 25 percent of revenue. The federal pipeline, which historically has lagged commercial adoption, accelerated meaningfully in 2025 after the Department of Defense added Vast to its joint warfighting cloud capability vehicle.
The international roster is expanding aggressively as well. Vast already operates in Saudi Arabia and the United Arab Emirates, and the Series F press release cited MENA expansion as a primary use of proceeds. The company’s positioning in the region benefits from existing relationships with Humain, the Saudi sovereign AI vehicle, and G42, which is building gigawatt-scale AI campuses in the UAE. India and Singapore are next on the geographic roadmap, with hires planned in both markets in the second half of 2026.
Vast Data vs Pure Storage, NetApp, Dell EMC, and IBM: A Storage Market Disrupted
The traditional enterprise storage market that Pure Storage, NetApp, Dell Technologies, and IBM have shared for two decades is being reshaped by AI workloads in real time. Vast’s $30 billion private valuation now exceeds Pure Storage’s public market capitalization of approximately $19 billion as of late April 2026 and rivals NetApp’s roughly $24 billion. The disparity is striking because Vast still ships substantially less revenue than either incumbent, but the public markets are pricing those incumbents on their legacy enterprise franchise while Vast is being priced on its AI growth curve.
| Vendor | 2025 Revenue | Valuation (April 2026) | Architecture | Largest AI Customer |
|---|---|---|---|---|
| Vast Data | ~$500M+ CARR | $30B (private) | DASE (NVMe over Fabric) | CoreWeave ($1.17B contract) |
| Pure Storage | $3.17B | ~$19B (public) | FlashArray, FlashBlade | Meta (FlashBlade) |
| NetApp | $6.57B | ~$24B (public) | ONTAP, AFF | NVIDIA (BlueXP) |
| Dell Technologies (storage) | ~$16.4B | Part of $77B parent | PowerStore, PowerScale | xAI (PowerScale) |
| IBM Storage | ~$3.4B | Part of $215B parent | FlashSystem, Spectrum | NVIDIA (DGX SuperPOD) |
| WekaIO | ~$200M (est.) | $1.6B (private) | Software-defined | Stability AI |
The competitive response has been belated but is now underway. Pure Storage launched its AI-focused FlashBlade//EXA platform in 2025, which extends the FlashBlade scale-out architecture to handle larger training workloads. NetApp has paired its ONTAP file system with NVIDIA’s BasePOD reference designs and signed an expanded reseller agreement with the GPU maker. Dell continues to differentiate through its bundled compute-and-storage AI Factory program, which packages PowerScale storage with PowerEdge GPU servers. IBM has the smallest AI footprint of the four and is leaning on its consulting business to defend share. None of those efforts have closed the technical gap on extremely large clusters, where Vast continues to win essentially uncontested.
The DASE Architecture: How Vast Won the AI Data Layer
The technical foundation of Vast’s success is a design choice the company made years before the AI boom. Traditional enterprise storage arrays bind compute and storage together inside a controller pair, which limits how many ports and how much CPU you can throw at a single dataset. Vast’s Disaggregated Shared Everything architecture instead places stateless compute nodes called CNodes on one side of an NVMe over Fabric network and dense flash enclosures called DBoxes on the other. Any CNode can read or write any byte on any DBox, which means the array scales near-linearly as you add either compute or capacity.
That separation matters enormously when a customer like xAI fires up tens of thousands of GPUs that simultaneously demand petabytes of training data. Legacy arrays force the customer to either shard the dataset across many appliances or accept that storage will throttle GPU utilization. Vast’s CNodes can grow to thousands of independent compute units serving the same logical namespace, eliminating the sharding problem entirely. The architecture also benefits from QLC flash, which is roughly half the cost per terabyte of TLC, by using the company’s similarity-based data reduction to push effective dollar-per-terabyte well below traditional NVMe arrays.
The architecture has now grown into what Vast calls a platform. The Vast AI Operating System layered on top includes the Vast Data Engine, which exposes serverless functions that run inside the storage fabric, and the Vast InsightEngine, which integrates an embedding pipeline so customers can vectorize unstructured data without moving it to a separate database. Both pieces are positioned as direct competition to standalone vector databases like Pinecone and Weaviate, and they are the layer that justifies Vast’s repositioning as an “AI operating system” rather than a storage product.
NVIDIA’s Strategic Investment: A GPU Maker Underwrites the Data Layer
NVIDIA’s continued participation in the Series F is the second most important detail in the round, behind the valuation itself. Jensen Huang’s company has been on Vast’s cap table since the Series E in late 2023, and the additional check signals that NVIDIA sees Vast as a strategic asset rather than a passive investment. Vast is one of the few storage vendors certified for the NVIDIA Cloud Partner program, the qualification framework that GPU clouds like CoreWeave, Crusoe, and Lambda must meet to resell NVIDIA reference designs. That certification effectively guarantees Vast a seat at the table whenever a hyperscaler or sovereign AI buyer specs a new GPU cluster.
The strategic logic for NVIDIA is straightforward. The biggest threat to NVIDIA’s GPU business is not AMD’s MI400 or Google’s TPU 8t but rather the bottleneck of feeding GPUs at the rate they can compute. If storage chokes, customers buy fewer GPUs. By underwriting the platform that consistently feeds the largest clusters, NVIDIA reduces friction in its own sales motion. The same logic explains the company’s investments in CoreWeave, Lambda, Nebius, and Wayve. Vast is the storage line item in the same playbook.
What is less clear is whether NVIDIA’s relationship with Vast is exclusive in any meaningful sense. NVIDIA has invested in DDN and continues to certify reference designs from Pure Storage, NetApp, Dell, and IBM. The competitive moat for Vast is therefore not the NVIDIA partnership itself but the architectural fit between DASE and the largest GPU clusters. The investment, in this reading, is recognition rather than insurance.
Market Context: The $76 Billion AI Data Infrastructure Wave
The Vast Data round lands in the middle of a broader buildout that is reshaping global capital markets. Big Tech’s hyperscale capital expenditure across Microsoft, Alphabet, Meta, Amazon, and Oracle is on track to clear $650 billion in 2026, a figure that reflects unprecedented spending on data centers, GPUs, networking, and the storage that ties them together. According to IDC, the AI infrastructure market grew 47 percent year over year in 2025, with storage representing roughly 12 percent of total spending. That implies an addressable market in the range of $76 billion in 2026, of which Vast captures only a low single-digit share today.
The runway argument is what investors keep coming back to. Even at $500 million in CARR, Vast holds less than 1 percent of total addressable storage spend in AI. If the market continues to grow at 30 percent annually for several more years and Vast maintains its current win rate against legacy arrays in the largest GPU clusters, the company can plausibly approach $5 billion in ARR by 2028. At a public market multiple of even 10 times forward revenue, that is a $50 billion company without heroic assumptions.
The risk argument is equally pointed. The same hyperscalers that drive the buildout, particularly Microsoft, Amazon, and Google, are also building their own first-party storage tiers tuned for their internal AI workloads. AWS’s S3 Express One Zone and Microsoft Azure’s Blob Premium tier compete directly with Vast for AI training data. If hyperscaler captive storage closes the architectural gap with Vast over the next 24 months, the addressable market for third-party AI storage could compress meaningfully.
The IPO Question: Why Vast Took $500 Million in Secondary First
The most-watched line item in the Series F is the secondary tender. Vast disclosed that more than half of the $1 billion raised was secondary capital, meaning new investors bought shares from existing holders rather than from the company itself. That structure is now standard for late-stage AI companies including OpenAI, Anthropic, and Cursor, all of which have run nine-figure tender offers in the last 12 months. The implications, however, vary widely.
For Vast, the tender accomplishes three things at once. First, it gives early employees and angel investors a path to liquidity without the company having to file an S-1, a meaningful retention tool when AI talent is being recruited aggressively by every neocloud and frontier lab. Second, it locks in a $30 billion benchmark valuation that the company can carry into IPO marketing. Third, it minimizes dilution at a moment when Vast is profitable and does not actually need primary capital to fund the business.
“Funding to prepare the company for a potential IPO,” reads the press release, the closest the company has come to confirming a public listing timeline. CEO Hallak has previously suggested the company could be IPO-ready by the end of 2025, a target that has now slipped into 2026 or 2027. The Series F is best understood as a pre-IPO clean-up round rather than a growth raise, and the secondary-heavy structure suggests an IPO is more likely 12 to 18 months out than imminent.
Industry Reaction: Analysts Call the Round a Watershed for Storage
Reaction across the storage and infrastructure analyst community has been pointed. “VAST has fundamentally rewritten the economics of large-scale AI storage,” Patrick Moorhead of Moor Insights & Strategy wrote in a research note following the announcement. “When you can ship at $500 million CARR with a 228 percent Rule of 40, you have either disrupted a market or invented one. In Vast’s case, both are true.” Camberley Bates of The Futurum Group added that Vast’s valuation “puts a number on what every CIO has been hearing from their AI architects for two years, namely that traditional dual-controller storage cannot keep up.”
Holger Mueller of Constellation Research framed the round in terms of the broader AI infrastructure market. “The Vast valuation tells you that storage is now a scaled software business, not a hardware one. The same shift happened in compute when Snowflake separated query from storage. Vast is the storage equivalent.” Mueller pointed to the company’s expanding software margin profile, which he expects to approach Snowflake’s 70 percent gross margin range as the AI Operating System layer matures.
The reaction from competitors has been more measured. Pure Storage chief executive Charles Giancarlo, on a recent earnings call, declined to comment on Vast specifically but emphasized that “the AI workloads we see in the field are increasingly diverse, and a single architecture is not going to win every customer.” NetApp chief George Kurian similarly downplayed competitive pressure, citing his company’s enterprise installed base as a structural moat. Both vendors are nonetheless expected to accelerate their AI-specific roadmaps in the second half of 2026 in direct response to Vast’s continued share gains.
Five Predictions for Vast Data and the AI Storage Market
The Series F is a moment to take stock of where the AI storage market is heading. Five plausible outcomes follow from the data Vast has now disclosed. First, Vast Data files for an IPO in the first half of 2027 at a target valuation between $40 billion and $60 billion, with the secondary-heavy Series F serving as the cap table cleanup that almost always precedes a public listing. The company has the metrics to support a top-quartile public market multiple, and the compounding dynamics of the AI buildout argue for a richer mark than today’s $30 billion.
Second, Pure Storage and NetApp both make defensive acquisitions in the AI storage space within 18 months. The most likely targets are WekaIO, Hammerspace, or one of the smaller object storage vendors. The strategic rationale is the same one that drove EMC to buy XtremIO in 2012, namely that it is faster to buy a modern architecture than to retrofit a legacy one. Third, NVIDIA expands its strategic investments to include at least one additional AI storage vendor as a hedge against Vast becoming a single point of dependency in the GPU supply chain.
Fourth, Vast Data crosses $1 billion in committed annual recurring revenue by the end of fiscal year 2026, doubling the current $500 million figure. The path runs through expansion of the CoreWeave and xAI accounts, federal program wins, and continued MENA growth. Fifth, the AI storage market consolidates around three architectures by 2028: Vast’s DASE design, the hyperscaler captive object stores, and a residual position for Pure FlashBlade in mixed enterprise environments. The remaining competitors either reposition around niche workloads or are absorbed.
What the Vast Data Round Means for Buyers and Builders
For enterprise buyers, the Series F is a signal that Vast Data has moved from emerging vendor to default option for large GPU deployments. Procurement teams that previously had to defend selecting an upstart over Pure Storage or NetApp can now point to a $30 billion valuation, profitability, and a customer list that includes the largest GPU clusters on the planet. The remaining decision becomes architectural fit, not vendor risk. Buyers should still negotiate hard on per-petabyte pricing, particularly for multi-year commitments, since Vast’s capacity to discount is constrained by the public valuation it now needs to defend.
For builders inside neoclouds and AI labs, the Vast platform’s Data Engine and InsightEngine layers are the most consequential parts of the announcement. The shift to embedded vector search and serverless compute inside the storage fabric closes a gap that previously required customers to assemble a separate vector database, message queue, and embedding pipeline. Vast is essentially betting that the data layer becomes the connective tissue of AI applications, with everything else collapsing into it. Whether or not that bet pays off, builders should evaluate the platform with the assumption that the surface area is expanding rapidly.
The Bottom Line: A $30 Billion Storage Company Is the New Normal
Vast Data’s $1 billion Series F at a $30 billion valuation is not the most expensive private financing of the AI era. SpaceX, OpenAI, Anthropic, and xAI all clear that bar. It is, however, the most expensive private financing for an infrastructure company that has historically been priced as plumbing rather than as an AI play. The implication is that the data layer underneath the GPU is no longer commodity, and the vendors who get the architecture right are going to be repriced into the same valuation tier as the model labs themselves.
The next 12 months will test whether the $30 billion mark is a ceiling or a floor. Vast’s revenue growth, the trajectory of its federal pipeline, and the competitive response from Pure Storage and NetApp will all weigh on the answer. What is no longer in dispute is that storage, after a decade of sleepy single-digit growth, has become one of the most consequential bottlenecks in the AI buildout. Whoever solves it owns a piece of every GPU dollar that ships, and Vast Data has positioned itself as the leading candidate.
Frequently Asked Questions
How much did Vast Data raise in Series F and at what valuation?
Vast Data raised approximately $1 billion in Series F funding at a $30 billion post-money valuation, announced on April 22, 2026. The financing included both primary capital and more than $500 million in secondary capital, more than tripling the company’s prior $9.1 billion Series E valuation from late 2023.
Who led the Vast Data Series F round?
Drive Capital and Access Industries co-led the round, with participation from existing investors NVIDIA, Fidelity Management & Research Company, and New Enterprise Associates (NEA).
What is Vast Data’s annual recurring revenue?
Vast Data disclosed more than $500 million in committed annual recurring revenue (CARR) at the close of fiscal year 2025, alongside more than $4 billion in cumulative bookings. The company is profitable and cash-flow positive, with revenue tripling year over year.
Who are Vast Data’s largest customers?
Named customers include CoreWeave (a $1.17 billion multi-year contract), xAI’s 200,000-GPU Colossus cluster, Lowe’s, the United States Air Force, and Cursor. The customer roster spans neoclouds, sovereign AI labs, federal agencies, and enterprise AI adopters.
Is Vast Data planning an IPO?
Vast has indicated that the Series F is intended in part to prepare the company for a potential IPO. The secondary-heavy structure, in which more than half of the $1 billion was sold by existing shareholders, suggests a public listing is more likely 12 to 18 months out than imminent. CEO Renen Hallak previously suggested the company could be IPO-ready by late 2025, a target that has slipped into 2026 or 2027.
How does Vast Data compare to Pure Storage and NetApp?
Vast Data’s $30 billion private valuation now exceeds Pure Storage’s roughly $19 billion public market capitalization and rivals NetApp’s roughly $24 billion, despite shipping less revenue than either. Both incumbents are pricing on legacy enterprise franchises, while Vast is being priced on its AI growth curve and its Disaggregated Shared Everything (DASE) architecture, which scales horizontally for very large GPU clusters.
What is Vast Data’s DASE architecture?
Disaggregated Shared Everything (DASE) is Vast’s storage design that separates stateless compute nodes (CNodes) from dense flash enclosures (DBoxes) over an NVMe over Fabric network. Any compute node can read or write any byte on any storage box, which lets the platform scale near-linearly without sharding data across appliances. The architecture is the technical foundation of Vast’s wins in extremely large GPU clusters.
Sources and further reading: Press release archive at VAST Data, market analysis from Tech Funding News, financial reporting from Globes, technical coverage from Blocks & Files, and additional context from TNW.
