Flow Traders, one of the more aggressive quantitative trading firms pushing into AI-driven market modeling, just made a decision that a growing list of hedge funds, research labs, and tech giants have already made before it. Rather than run its AI training workloads on a general-purpose cloud platform, the firm is handing that job to CoreWeave, the Livingston, New Jersey-based company that has quietly become one of the most important names in the entire AI infrastructure economy.
The arrangement, announced this year, makes CoreWeave the primary AI cloud platform provider for Flow Traders’ recently formed AI and deep learning division. Under the deal, Flow Traders will shift its most intensive training workloads onto CoreWeave Cloud, gaining dedicated compute capacity built specifically to train the foundation models at the heart of its AI-driven quantitative trading strategy. It’s a significant vote of confidence from a firm whose entire business depends on split-second decision making, and it says something about how far CoreWeave has traveled since its earliest days as a scrappy crypto mining operation running out of New Jersey.
What CoreWeave Actually Is, and Why It Isn’t Just Another Cloud Company
To understand why a firm like Flow Traders would choose CoreWeave over Amazon Web Services or Google Cloud, it helps to understand what CoreWeave actually does differently. Most cloud providers are built to be generalists, offering a bit of everything to every kind of customer, from small businesses hosting websites to enterprises running databases. CoreWeave was built around a much narrower and more demanding premise: a “GPU-first” model, in which entire data centers are engineered from the ground up around high-performance graphics processing units supplied by Nvidia, rather than treating GPU capacity as one option among many.
That distinction matters enormously to the kind of customer CoreWeave actually serves. The company doesn’t chase broad market share the way a traditional cloud provider does. It targets top-tier AI labs, deep learning startups, and technology giants, companies like OpenAI, Meta, and Anthropic, that need massive fleets of servers to train large language models and run heavy AI inference workloads around the clock. Those workloads don’t tolerate the kind of latency and bottlenecking that can slip by unnoticed in a general-purpose cloud environment, which is why CoreWeave has built its infrastructure around hyper-fast specialized networking and storage, running on Kubernetes, specifically engineered to move terabytes of data without the friction that slows down traditional cloud architecture.
The Nvidia relationship is central to all of it. CoreWeave secures large quantities of the most sought-after AI chips on the market, including H100s and newer-generation clusters, and because Nvidia is itself a major investor in CoreWeave, the company frequently gets access to hardware that other well-funded startups simply can’t secure at the same scale or speed. In an industry where chip availability has become one of the single biggest constraints on how fast a company can grow, that access is close to a structural advantage.
From Ethereum Mining Rigs to the Nasdaq
CoreWeave’s origin story reads like a case study in pivoting at exactly the right moment. The company began in 2017 as Atlantic Crypto, founded in Livingston, New Jersey by three former commodities traders, Michael Intrator, Brian Venturo, and Brannin McBee. Their original business had nothing to do with artificial intelligence. They were buying enormous quantities of GPUs to mine Ethereum, riding the cryptocurrency wave that had traders and technologists alike racing to accumulate computing hardware.
Then the crypto market cratered in 2018, and the founders were left holding a massive fleet of GPU hardware that suddenly had far less obvious value in the crypto space that had justified buying it. Rather than liquidate, they recognized that the same compute-heavy hardware could serve an entirely different, and ultimately much larger, market: enterprise and research workloads that needed exactly the kind of raw processing power sitting in their data centers. By 2019, the company had rebranded as CoreWeave and repositioned itself entirely around cloud computing services, trading crypto mining for the infrastructure business that would eventually make it one of the most closely watched companies on Wall Street.
The scale of that transformation became undeniable once the generative AI boom took hold. CoreWeave went public in March of 2025, listing on the Nasdaq under the ticker CRWV, and the growth since then has been difficult to overstate. Multibillion dollar compute infrastructure agreements with clients including Meta and OpenAI have pushed the company’s total forward revenue backlog past $99 billion, a figure that places CoreWeave among the most heavily contracted infrastructure providers in the entire AI sector, despite having existed as a cloud computing company for barely half a decade.
Why Flow Traders Chose CoreWeave
The Flow Traders agreement fits neatly into a broader pattern CoreWeave has pointed to directly: quantitative trading firms are increasingly building out dedicated AI and deep learning divisions, treating the training of foundation models not as an experimental side project but as a core competitive discipline in its own right. That shift changes the infrastructure requirements dramatically. Training frontier-scale models for financial market prediction demands consistent multi-node performance at a scale that most general-purpose cloud platforms simply weren’t designed to sustain, which is precisely the gap CoreWeave has built its business around closing.
Joshua Mathew, co-head of Flow Traders’ newly announced AI and deep learning division, framed the decision in blunt terms, arguing that tomorrow’s innovations cannot be built on yesterday’s infrastructure and that the firm’s research into modeling financial markets required the technology and scale CoreWeave could provide. According to the companies, Flow Traders arrived at that conclusion only after a competitive evaluation process that weighed multi-node performance, technical support, and long-term roadmap planning, with CoreWeave’s ability to sustain performance under large-scale, high-intensity training workloads emerging as a decisive factor.
Jon Jones, CoreWeave’s chief revenue officer, described Flow Traders as one of the financial services firms actively pushing the boundaries of what’s achievable with AI-driven trading, a category of customer that requires infrastructure capable of holding steady under multi-node, high-intensity demand without buckling. CoreWeave has positioned its entire platform around exactly that promise: combining cluster-level performance and reliability with the software and support layer needed to help customers move beyond model training and into live production applications and autonomous agents.
The Credibility Behind the Pitch
CoreWeave’s case for why firms like Flow Traders should trust it with mission-critical training workloads doesn’t rest on marketing language alone. The company has pointed to a run of third-party performance validation to back up its claims, including strong results in MLPerf benchmarking, Platinum-tier rankings in SemiAnalysis’s ClusterMAX evaluations across both its 1.0 and 2.0 assessments, and independent inference benchmarking conducted by Artificial Analysis for Moonshot AI’s Kimi K2.6 model. In an industry crowded with infrastructure providers making similar performance claims, that kind of external, repeatable validation has become one of the clearer ways to separate genuine capability from marketing copy, and it’s part of why a firm as demanding as Flow Traders was willing to shift its most sensitive workloads onto CoreWeave’s platform.
A New Jersey Company at the Center of the AI Infrastructure Race
There’s something worth sitting with in the fact that a company now central to how OpenAI, Meta, Anthropic, and now a major European trading firm train their most advanced models started less than a decade ago as a crypto mining operation in a New Jersey office park. CoreWeave’s rise mirrors the broader AI infrastructure buildout happening across the industry, where the physical constraints of chip supply and data center capacity have become just as important to the AI race as the algorithms themselves. As more firms outside the traditional tech sector, from quantitative trading houses to research institutions, come to the same conclusion Flow Traders reached, CoreWeave’s position at the center of that buildout looks less like a temporary advantage and more like the foundation of a genuinely new category of infrastructure company, one that happens to trace its roots back to a small New Jersey suburb and a bet on Ethereum that didn’t pay off the way its founders originally expected.
