KONST Closes $30M Series B Led by ADATA Technology
TREE NEWS reports: Taiwan-based AI compute operator KONST has completed a $30 million Series B funding round led by memory and storage manufacturer ADATA Technology, with strategic participation from M Mobility and Pegatron Venture Capital. The company said proceeds will be used to expand its AI data center footprint across Asia and accelerate its “Token Factory” strategy.
A Three-Layer Stack: Compute, Metering, Governance
KONST’s architecture is organized into three layers, each addressing a distinct bottleneck in enterprise AI adoption:
- Konstra AI (base layer): Supplies raw compute and operates the underlying data centers.
- Glows.ai (middle layer): Breaks cluster capacity into small, discrete units and bills by the second, enabling granular, on-demand consumption rather than long-term reservations.
- Horizon AI (top layer): Issues the (k) ATP Token, which handles governance and delivery, offering what the company describes as one-click access to all models while unifying enterprise AI permissions, quotas, and audit logs.
Why Tokenized Compute Matters
The (k) ATP Token is the most interesting piece of the announcement. By tying governance and access entitlements to a token, KONST is effectively attempting to build a settlement and rights-management layer for AI inference — a market that has historically been priced opaquely through cloud contracts and reserved instances. If the token genuinely governs quota allocation and audit trails, it creates a verifiable paper trail for enterprise AI spend, a pain point for compliance teams at large organizations.
This places KONST in a growing cohort of firms converging AI infrastructure with crypto-native rails. Decentralized GPU networks have spent years trying to aggregate idle compute, but enterprise-grade demand has largely stayed with hyperscalers because of reliability, latency, and legal guarantees. KONST’s approach is notable for being enterprise-first: instead of sourcing fragmented supply from anonymous providers, it builds and operates its own data centers and then tokenizes the access layer on top.
Asia’s AI Infrastructure Race
The investor list is telling. ADATA brings memory and storage supply chain depth; Pegatron Venture Capital connects to one of the world’s largest electronics manufacturing ecosystems. Taiwan sits at the center of the AI hardware supply chain, and a domestic compute operator with backing from storage and manufacturing players is a bet on regional sovereignty over AI capacity — a theme gaining traction as export controls and data residency rules reshape where models can be trained and served.
Forward Look
The key questions for KONST are execution and adoption. Per-second billing is attractive in theory but requires utilization high enough to cover fixed data center costs. The (k) ATP Token must demonstrate real governance utility rather than functioning as a loyalty point. And the company will need to prove that tokenized access can satisfy enterprise procurement, security review, and audit requirements — the real gatekeepers of AI budgets.
If KONST can convert its supply chain relationships into anchor enterprise customers, it could establish a template for how AI compute is packaged, priced, and settled in Asia. That would make the Series B less about capacity expansion and more about defining a new market structure.
