KONST announces the completion of a $30 million Series B financing round, led by ADATA Technology (TPEx: 3260), with participation from M Mobility, Pegatron Venture Capital, and several other corporate investors.

This next-generation AI computing power operator, headquartered in Taiwan, stated that it will use this funds to expand the construction of its AI data centers in Asia and to accelerate the advancement of its Token Factory strategy in order to meet the growing adoption needs of rapidly developing enterprises AI.

Series B financing: Strategic participation from multiple corporate groups

The main investors in this round come from the fields of storage and flash memory, electrification solutions, and electronic manufacturing. Other participants cover industries such as energy, semiconductors, and thermal management – which are also sectors that have a demand for AI. KONST stated that, in addition to the capital itself, the company values more the synergies that investors can bring in terms of industry application scenarios and customer applications, which will help to connect the company's services with actual deployments more quickly.

The significance of this round of financing is not just about raising a certain amount of capital. Most importantly, we share the same vision with our investors: AI must truly be integrated into the actual operations of the company, rather than remaining at the proof-of-concept stage, said KONST, co-founder and chairman of Ben Chang. “Once AI becomes part of the operations, computing power is no longer a one-time expense, but an infrastructure that the company relies on every day. Our investors understand this better than anyone, as they are also part of the demand side themselves.”

From data centers to governance in use: The three-layer architecture of KONST transforms computing power into services

KONST indicates that enterprises' demand for AI is shifting from pilot projects involving a single team to regular, cross-departmental use; accordingly, the demand for computing power has also shifted from one-time project purchases to continuous, fluctuating daily consumption. Fixed capacity configurations are increasingly unable to keep up with this changing pattern of usage. At the same time, as agentic and AI gradually mature, workloads have higher requirements for computational performance, stability, and reliability. With both the scale and quality of demands increasing, this is precisely why AI infrastructure becomes even more important.

Token Factory views computing power as production equipment, and token as its output. KONST is advancing this concept through a three-tier architecture:

At the underlying level is the supply of computing power. Konstra AI is responsible for everything from site selection, machinery, electrical systems, and environmental control system construction, to GPU cluster tuning and acceptance, to subsequent operation and maintenance. The company stated that the funds raised in this round of financing will be used to expand its construction capabilities in Asia.

The intermediate layer is the conversion layer. Glows.ai divides the cluster's computing power into smaller units, allowing enterprises to be billed per second, paying only for the actual amount of usage.

At the top level is governance and delivery. The (k) of Horizon AI and ATP Token are responsible for this layer, deciding who can use what, how much to use, and how to charge for it.

KONST believes that this three-tier division of labor also reflects how Taiwan's existing hardware and supply chain advantages can be further extended upstream in the value chain – from manufacturing computing power to delivering computing power.

(k) ATP Token: One key to access all models and manage team usage

Governance is the most easily overlooked layer among these three, and it is also where companies often get stuck. The problem is not with insufficient model capabilities, but rather with the lack of clarity around which departments use which models, how much money is spent on them, and what the results of each request are.

(k) ATP Token is designed around the principle of "one key, all models." Enterprises only need one project key to connect to all authorized models, without having to negotiate contracts with each supplier individually or perform integration work for each one.

The platform adopts a four-layer structure: organization, workspace, project, and key. Quotas can be allocated at the organizational level, model access permissions can be set at the project level, and alerts are issued before budget overruns occur. Each request's used models and quantities are recorded in the audit logs. For enterprises that are already operating on a large scale, (k) ATP Token goes a step further by consolidating all models on the platform into a single contract and bill, and provides customized contract pricing based on actual usage. KONST emphasizes that (k) ATP Token is not an independent product, but rather a governance foundation embedded within each project using AI. From the very first day of proof of concept, cost tracking, permission management, and request logging were in place; therefore, going live is merely a matter of scaling up traffic.

“Buying GPU is just the beginning,” said Chang. “You also need a buildable data center, divisible computing power, and a controllable way of using it. Only by connecting these three elements can we provide what we call computing power delivery.”

Focusing on the AI adopted by large enterprises

KONST indicates that Horizon AI is primarily targeted at large corporate groups and large organizations. The AI of such customers typically involves multiple departments, integration with existing systems, and long-term governance requirements. What they need is not a point solution, but partners who can collaborate throughout the entire adoption cycle. Horizon AI has already completed deployments for numerous enterprises and institutions and is currently conducting requirement assessments with major clients in various industries.

As the business grows, KONST is also continuously expanding its team by recruiting talents from various fields to meet the increasingly diverse needs of customers. For recruitment information, please visit the LinkedIn page on KONST: https :// www.linkedin.com / company / konsttech.

About KONST

KONST Co., LIMITED is a next-generation AI computing power operator headquartered in Taipei, Taiwan. Through the One KONST brand and its strategic partner ecosystem, KONST integrates the capabilities of three major brands: Konstra AI, Glows.ai, and Horizon AI, providing a complete AI computing power value chain in the form of integrated services. Konstra AI offers AI data center computing power construction services, covering end-to-end construction from mechanical, electrical, and environmental control systems to GPU clusters, while also providing equipment operation and maintenance as well as computing power delivery. Glows.ai provides enterprise-level GPU cloud platforms with billing based on seconds. Horizon AI focuses on the adoption of AI applications and model asset governance. Customers only need to collaborate with one team and place one order, regardless of whether their needs lie at the computing power layer, platform layer, or application layer.

Leveraging Taiwan's engineering strength and supply chain advantages, KONST integrates the operations of GPU clusters, data center maintenance, cloud platforms, and AI services. Amidst the global AI trend, the company collaborates with enterprises, developers, research institutions, and international partners to build a scalable, reliable, and sustainable computing power supply.

Related Links

(k) ATP Token, Horizon AI's enterprise AI governance platform: https :// www.horizon-ai.ai / en / atp

(k) ATP Token Console: https :// atptoken.ai /

Media Contact

KONST Co, LIMITED