KONST says $30 million Series B will fund more AI data centers in Asia

KONST, a Taipei-based AI compute operator, says it has raised $30 million in Series B funding led by ADATA Technology, a Taiwan-listed memory and storage company. The company announced the round on Oct. 9, 2026, and named M Mobility and Pegatron Venture Capital among several other corporate investors. This article rests on that announcement alone. We found no independent confirmation of the amount, the investors or the plans, so every figure below is the company's own account.

What is new is the stated purpose of the money. KONST says it will expand its AI data-center buildout across Asia and push what it calls a "Token Factory" strategy, which treats computing hardware as production equipment and AI tokens as the output. The release gives no site locations, capacity figures, valuation or timeline for the buildout.

The company describes three layers. Konstra AI handles data-center projects from site selection and electrical and cooling systems through GPU cluster tuning and operations. Glows.ai divides cluster capacity into smaller units sold with per-second billing. Horizon AI's (k) ATP Token sits on top, controlling who may use which models, how much, and how usage is charged. KONST says the platform uses a four-tier hierarchy of organization, workspace, project and key, records each request's model and usage in audit logs, and lets large customers put every model under one contract and invoice.

Confirmed, in the narrow sense, is only that KONST issued this statement through PR Newswire. Not verified are the round size, the identities and stakes of the investors, and the company's report that Horizon AI has finished deployments for several enterprises and is assessing needs with large customers; no customers are named. The investors are described as coming from memory, electrification, electronics manufacturing, energy, semiconductors and thermal management, and KONST frames them as sitting on the demand side of enterprise AI.

It matters because the pitch ties physical capacity to metering and governance. Enterprises that move AI from pilots to daily use must account for spending and access by department, and KONST argues that this, not model quality, is where adoption often stalls. Whether its approach beats rival cloud and model-access offerings is untested in the material available. Readers should treat the claims as a company's plan until customers, capacity or regulatory filings are published.

Revised 16 October 2026: Guth re-read the original sources and rewrote this article. The earlier version remains available as revision 1.