Datavault AI (NASDAQ: DVLT) completed a $60 million registered direct offering on May 5, selling 109,090,910 shares of common stock at $0.55 each. The proceeds fund deployment of a quantum-ready GPU edge network across approximately 100 U.S. cities. The stock dropped 25.5% on the announcement.
The Offering
The company announced the deal was led by institutional investors with Titan Partners, a division of American Capital Partners, acting as sole placement agent. The shares were filed under a shelf registration statement on Form S-3 that the SEC declared effective in March 2026.
CEO Nathaniel T. Bradley called the financing “an important step in the deployment of our quantum-ready GPU edge network,” adding that the capital would “position Datavault AI to capture growing demand for AI infrastructure.”
$180M Total Capital Commitment
The $60M offering follows a separate binding term sheet announced April 27 with Scilex Holding Company for a $120 million cash contribution. Under that agreement, Scilex will contribute $120M by December 31, 2026, in exchange for a portion of certain revenues generated from the GPU edge network.
Combined, the two deals give Datavault AI $180 million to build out its SanQtum AI platform, which distributes GPU compute across metropolitan markets rather than concentrating it in centralized data centers.
The Dilution Problem
The market reaction tells its own story. According to Simply Wall St, DVLT fell 25.5% after the announcement, reflecting investor concerns about shareholder dilution layered on top of execution risk. The company remains loss-making, and the central question is whether $180M in infrastructure capital can convert into margin-accretive revenue before the cash runs out.
Edge Compute in a Centralized World
Datavault AI’s distributed model sits at an interesting crossroads in the AI infrastructure market. The dominant trend in 2026 has been centralized megacompute: Anthropic committed $200 billion to Google Cloud over five years, SpaceX’s Colossus facility runs 220,000 GPUs at 300MW, and Project Stargate is building out at even larger scale.
Datavault is betting the opposite direction. Its SanQtum platform, operated on Available Infrastructure’s network, pushes GPU capacity to the edge for inference workloads, real-time analytics, and enterprise computing across financial services, sports, media, and life sciences. According to Pulse 2.0, the company’s architecture “differentiates its offering from conventional data center models by distributing compute capacity across metropolitan markets.”
Whether distributed edge GPU networks can compete with hyperscale centralization for AI workloads remains an open question. The $180M war chest gives Datavault roughly 18 months to prove the thesis.