A Las Vegas-based cloud computing startup just broke Nevada’s record for the largest Series B funding in state history, and it did it by refusing to use a single Nvidia chip. TensorWave, a Las Vegas-based AI cloud provider that built its entire business around AMD chips, closed a $350 million Series B round that values the company at $1.55 billion, roughly tripling its implied valuation from just over a year ago. TensorWave challenges Nvidia not with incremental competition, but with a clean-sheet, all-AMD strategy — and a growing roster of enterprise customers already running production workloads on it.
Less than a month after TensorWave was founded in December 2023 by Darrick Horton, Piotr Tomasik, and Jeff Tatarchuk, it received its first investment. Since then, the company has executed one of the most aggressive scaling trajectories among all gpu cloud computing startups in North America. AMD Ventures and Magnetar Capital co-led the Series B round, with participation from Maverick Silicon, Nexus Venture Partners, and Western Frontier, bringing total funding to approximately $493 million since founding.
Breaking the Nvidia AI Monopoly: Why It Matters Right Now
The market context behind breaking nvidia ai monopoly is stark. The AI chip market reached $120 billion in revenue in 2025 — triple 2023 levels — while Nvidia controls approximately 80% of the AI training chip market. That concentration creates pricing power, supply bottlenecks, and ecosystem lock-in that enterprises are increasingly motivated to escape. At peak demand in 2024, customers faced 6–12 month wait times for Nvidia H100 GPU orders — a severe bottleneck for AI development. For companies racing to move AI from experimentation into production, a six-month queue isn’t an inconvenience. It’s a competitive liability.
TensorWave’s 28-year-old CEO Darrick Horton landed on Forbes 30 Under 30 for AI for what Forbes called a “seemingly impossible mission”: breaking Nvidia’s dominance in AI infrastructure. Before founding TensorWave, he was working on plasma physics for nuclear fusion at Lockheed Martin’s Skunk Works — the division responsible for the U-2 spy plane and the SR-71 Blackbird. He brought that level of precision and purpose to a different kind of mission. Horton has said that when TensorWave started in 2023, Nvidia was a monopoly by default — rather than one built on anticompetitive practices — and that customers were eager to diversify into alternative ai chip suppliers. He also noted that he does not like buying things from monopolies because customers lack leverage.
Founded in 2023, TensorWave positions itself as an alternative to Nvidia, whose GPUs dominate AI infrastructure. “We were created to restore competition to the market,” CEO Darrick Horton said. Breaking nvidia ai monopoly runs through every decision the company makes — from chip selection to commercial strategy.
The All AMD AI Cloud That Nvidia Didn’t See Coming
What separates TensorWave from other gpu cloud computing startups is absolute commitment. There’s no Nvidia fallback, no hybrid stack, no hedged bets. TensorWave is the all AMD AI cloud specializing in high-performance, memory-intensive workloads, running exclusively on AMD Instinct Series GPUs paired with AMD’s ROCm open software stack. That single-vendor strategy creates deep product integration and, critically, access to alternative ai chip suppliers that aren’t constrained by Nvidia’s supply chain.
Since raising its Series A financing in May 2025, TensorWave has expanded significantly and now operates one of North America’s largest AMD-based AI training clusters, with 8,192 AMD Instinct MI325X GPUs online. TensorWave has also secured more than 2 gigawatts of long-term data center capacity — enough energy to power a small city — allocated to enterprise, research, and AI-native customers. The company already runs three AI data centers in the United States, located in Pennsylvania, Arizona, and Florida.
Customers including Fireworks AI and Luma AI are already using TensorWave’s infrastructure to support large-scale generative AI workloads and production inference systems. These aren’t test environments. They’re production deployments — the most credible validation an all AMD AI cloud can offer.
Why Memory-Intensive AI Workloads Are TensorWave’s Sharpest Edge
TensorWave challenges Nvidia on technical grounds just as much as commercial ones. AMD’s Instinct GPUs carry significantly more High Bandwidth Memory than comparable offerings, and that advantage is decisive for specific, fast-growing workload categories. The AMD Instinct MI300X GPU features 192GB of HBM3 memory and a bandwidth of 5.3 TB/s, purpose-built to handle the most demanding AI workloads. The AMD Instinct MI355X takes a massive leap forward with 288GB of HBM3E memory and 8TB/s of memory bandwidth, unlocking new performance ceilings for large model training and high-throughput inference.
For memory intensive ai workloads — large context window language models, real-time video generation, multi-modal foundation models, and high-volume inference pipelines — those specifications aren’t just competitive. They’re architecturally superior for the right use cases. Tomasik noted that “workloads that lend themselves to AMD better, like video workloads and large context workloads,” are specifically “things that need a lot of memory.” Those happen to be the fastest-growing categories in enterprise AI today.
The new capital will support the deployment of next-generation AMD Instinct MI355X GPU clusters designed for memory-intensive workloads such as large language model training, high-throughput inference, and generative AI applications. This focus on memory intensive ai workloads isn’t a niche play. It’s a deliberate bet on where the frontier of model development is heading.
A Funding Trajectory That Breaks Every Record
TensorWave’s fundraising history reads like a launch sequence. The company previously raised the largest startup funding in Nevada history in October 2024, landing $43 million in seed funding. Less than a month after TensorWave was founded in December 2023, it received an infusion of $2.2 million from StartUpNV. Then the climb steepened fast. TensorWave broke a record for the largest Series A funding secured in Nevada history, raising $100 million, led by Magnetar and AMD Ventures, with support from Maverick Silicon, Nexus Venture Partners, and new investor Prosperity7. The $350 million Series B now makes it three consecutive Nevada funding records for the same company.
The deal is structurally significant as it sees AMD — the primary chip supplier for TensorWave — using its own balance sheet to aggressively fund a major customer, mirroring a strategic playbook long utilized by Nvidia to secure market share for its own accelerators. AMD’s direct investment through its venture arm signals that AMD views TensorWave as more than just a customer, but as a strategic channel for proving that AMD silicon can compete at scale in production AI workloads.
In February 2026, TensorWave established a partnership with Credo, a networking technology company, to improve network reliability for large-scale GPU clusters. That move signals that TensorWave is building enterprise-grade reliability, not just raw compute density.
TensorWave Challenges Nvidia in a Market Built for Disruption
TensorWave isn’t fighting this battle alone. Other companies are also working to provide AI labs and large enterprise customers with more computing alternatives beyond Nvidia, including Cerebras, which makes platter-size chips for running AI models quickly, and Majestic Labs AI, which produces chips with large amounts of memory. These gpu cloud computing startups are collectively building an ecosystem of alternative ai chip suppliers with real production credibility.
The global cloud computing market was valued at $752.44 billion in 2024 and is projected to grow at a CAGR of 21.2% through 2030. The AI infrastructure subset is growing far faster: Nvidia’s own projections put the total AI chip and infrastructure market at $1 trillion by 2027. Most of that spend currently flows to Nvidia — but enterprises are increasingly aware that concentrating this critical resource with a single provider is a structural risk, not just a procurement preference.
Most AI infrastructure providers are capacity-constrained on Nvidia hardware, creating a supply bottleneck that AMD-based alternatives can fill. AMD Ventures has consistently backed this thesis and co-led Vultr’s $333M raise at a $3.5B valuation, another GPU cloud provider building on AMD infrastructure. The market is building depth fast.
AMD Instinct GPU Clusters: The Technical Foundation
The amd instinct gpu clusters underpinning TensorWave’s infrastructure run on AMD’s ROCm open-source software stack — a deliberate counter to Nvidia’s proprietary CUDA ecosystem. The company has worked to improve its ROCm software, making it easier to use, which helps AMD compete better with Nvidia’s popular CUDA platform. TensorWave’s engineers work directly with customers to tune workloads for AMD hardware, closing the workflow friction that once made CUDA the unquestioned enterprise default.
The company is preparing larger deployments of MI355X clusters across new data center regions in North America. Those amd instinct gpu clusters will extend TensorWave’s capacity into new geographies and workload categories, bringing the all AMD AI cloud model to a wider range of enterprise customers who currently run memory intensive ai workloads on constrained or overpriced Nvidia infrastructure. Since its founding, TensorWave has increased its overall capacity tenfold each year, and Horton has publicly said he intends to do so again.
What Comes Next: Scale, Hiring, and Ambition at Every Level
TensorWave is continuing to invest in its Las Vegas headquarters and expects to double the number of employees over the next 12 months, from about 160 to between 300 and 400. Roles span engineering, infrastructure, operations, sales, and customer success — the full organizational stack required to operate a global GPU cloud at scale.
Horton plans to have 100,000 GPUs deployed by next year, with hopes to eventually become the first company to deploy 1 million GPUs — a feat no company has ever achieved. These are genuinely ambitious targets. But every milestone TensorWave has set so far, it has hit ahead of schedule. At the current growth rate, skepticism may be the riskier position.
While the initial deployments in 2025 and early 2026 are focused on the 10-to-20 megawatt range, the long-term vision is to provide a sovereign AI cloud that can support the training of models with trillions of parameters. That’s not a startup roadmap. That’s an infrastructure ambition at the scale of national significance.
Conclusion: A Credible Challenger with Real Momentum
TensorWave challenges Nvidia not by matching it GPU-for-GPU, but by owning the market segment that Nvidia’s dominance has left under-served: open-ecosystem, memory-optimized, immediately accessible compute for enterprises that cannot afford to wait six months for hardware or be perpetually dependent on one vendor.
With $493 million in total funding, more than two gigawatts of secured data center capacity, production customers already running demanding memory intensive ai workloads, and amd instinct gpu clusters actively expanding across North America, TensorWave has moved well beyond challenger status. It’s now infrastructure. The question is no longer whether alternative ai chip suppliers can build something competitive. TensorWave has already answered that.
The global AI infrastructure market is heading toward $1 trillion by 2027. Nvidia will capture a large portion of it. But the era of a single company holding all the leverage is ending — and TensorWave is among the companies ending it.
Frequently Asked Questions
What exactly is TensorWave and what does it do?
TensorWave is a Las Vegas-based AI cloud provider founded in December 2023 that operates exclusively on AMD Instinct GPUs. It delivers high-performance, memory-optimized GPU cloud infrastructure for enterprise AI workloads — including large language model training, high-throughput inference, and generative AI applications — without using any Nvidia hardware.
How much has TensorWave raised, and who led the Series B?
TensorWave raised $350 million in a Series B round co-led by AMD Ventures and Magnetar Capital, with continued participation from Maverick Silicon, Nexus Venture Partners, and Western Frontier. The round values the company at $1.55 billion and brings total funding to approximately $493 million since founding.
Why does TensorWave use AMD chips instead of Nvidia?
TensorWave’s founding premise is to restore competition in a market dominated by a single supplier. CEO Darrick Horton has stated that Nvidia’s near-monopoly strips customers of leverage and creates supply bottlenecks. AMD’s Instinct GPUs also carry significantly more HBM memory — up to 288GB in the MI355X — making them technically superior for memory-intensive AI workloads like large context models and video generation.
Who are TensorWave’s real-world customers?
Confirmed customers include Fireworks AI and Luma AI, both of which are using TensorWave’s AMD-powered infrastructure for production-scale generative AI and high-throughput inference workloads — not pilots or proofs of concept.
Who founded TensorWave, and what is Darrick Horton’s background?
TensorWave was co-founded by CEO Darrick Horton, President Piotr Tomasik, and Chief Growth Officer Jeff Tatarchuk. Horton, 28, is a Forbes 30 Under 30 honoree who previously worked at Lockheed Martin’s Skunk Works on nuclear fusion research and holds degrees in Mechanical Engineering and Physics from Andrews University.
What AMD GPU hardware does TensorWave currently run?
TensorWave currently operates one of the largest AMD-based AI training clusters in North America, with 8,192 AMD Instinct MI325X GPUs deployed. The company is actively preparing next-generation AMD Instinct MI355X GPU cluster deployments across multiple new North American data center regions.
What are TensorWave’s growth plans following the Series B?
TensorWave plans to use the $350 million to expand its global AI infrastructure footprint through new MI355X deployments, grow its team from approximately 160 to between 300–400 employees over the next 12 months, and reach 100,000 GPUs deployed within the next year. The company has secured more than 2 gigawatts of long-term data center capacity and maintains a long-term ambition of becoming the first company ever to deploy one million GPUs.