TensorWave
Bottleneck owner · US
All-AMD AI cloud purpose-built for memory-intensive training/inference — an open-ecosystem alternative to Nvidia-locked neoclouds.
AMD Instinct MI300X/MI355X GPU clusters + high-performance inference engine.
Research mapping
Physical AI / AI Factory & Cloud Training Infrastructure (L1) / Bottleneck owner
Evidence boundary
Original analysis and independent corroboration remain under development. Commercial scale and durable technical advantage remain under review.
What makes it defensible
AMD-only AI cloud: MI300X/MI355X clusters plus a tuned inference engine, positioned as the credible non-Nvidia path to cheaper AI compute per token.
Position in the stack: Neocloud layer between accelerator supply and AI teams. The defensible part is contracted AMD accelerator allocation plus site and power execution; the software stack is a layer on top, not the moat.
Who buys it
Named customers: Fireworks AI, Luma AI
Founding team
Darrick Horton — Co-founder & CEO — Co-founded TensorWave in Las Vegas in 2023; previously co-founded crypto-mining company VaultMiner Technologies, parent of VMAccel.
Piotr Tomasik — Co-founder, President & COO — Co-founded TensorWave in 2023; previously co-launched influencer marketing platform Influential and Lets Rolo.
Jeff Tatarchuk — Co-founder & Chief Growth Officer — Co-founded TensorWave in 2023.
SourceFinancing detail
- Last round
- Series B ($350M @ $1.55B, Jun 2026)
Investors: AMD Ventures, Magnetar, Maverick Silicon, Nexus Venture Partners, Western Frontier
Mapped funds and investors
Direct relationships supported by published evidence.
Reported holdings as-of
No sourced relationships mapped yet.
Disclosed investment history
No sourced relationships mapped yet.
Showing up to four relationships in each group. ETF weights describe the fund’s portfolio, not ownership of this company. Private investment history does not establish current ownership.
Evidence & media
-
Fermi America announces $6.5 billion contract with TensorWave
What this source supports: Fermi Inc. first binding customer lease at Project Matador (Carson County, TX): TensorWave TEX1, LLC takes 222 MW of facility power for 15 years, ~$6.5B of contracted revenue, for tens of thousands of next-gen AMD Instinct GPUs; expansion rights to >650 MW across two further data centers.
What it does not establish: Lease is signed but capacity is not yet delivered; revenue is spread over 15 years and contingent on the facility being built and powered. No per-MW economics disclosed.
Source -
TensorWave powers frontier AI growth for cloud customers with AMD Helios rackscale solution
What this source supports: TensorWave adopts AMD Helios rackscale (72x Instinct MI455X GPUs, 6th Gen EPYC, Pensando networking, ROCm) as a rack-scale building block for frontier inference/training. AMD quotes a TensorWave customer (Featherless) on the choice.
What it does not establish: First-party/vendor announcement; no deployed rack count, utilization or economics.
Source