Singaporean startup Acrab has raised $350 million to build infrastructure for agentic AI.
Singapore-based company Acrab has announced raising over $350 million in total funding. These funds are aimed at creating a next-generation computing infrastructure focused exclusively on agentic artificial intelligence systems. This volume of investment is a clear market signal: agentic AI is becoming not just a trend, but a fundamental direction requiring specialized hardware and software.
Founded in 2024, Acrab has quickly developed an ambitious roadmap. The company is building a full-stack computing architecture: from its own AI chips to operating systems and multimodal interfaces. Special attention is given to local execution of large language models (LLMs) — a key element for tasks where processing speed and data privacy are critical.
GΞLIX Platform: From Tests to Industrial Deployment
Acrab's flagship product is the GΞLIX platform, designed for local LLM execution in agentic AI scenarios. According to the team, the solution has already been tested in real-world conditions and is in the final stage before its first industrial deployment and mass production launch. This means Acrab is targeting not laboratory experiments, but commercially mature solutions capable of competing with major players.
The funds raised will be used to scale the platform, deepen research in computing systems, expand the partner network, and enter international markets. For the crypto community, there is an important nuance here: the development of specialized AI chips and local LLMs could drive demand for decentralized computing resources and tokenized computing power — precisely the niche being actively explored by projects in the DePIN (Decentralized Physical Infrastructure Networks) space.
My expertise: Acrab has chosen the right strategy by focusing on the niche but rapidly growing segment of agentic AI. However, the key challenge will be not so much development as integration with existing blockchain ecosystems. If Acrab can offer a transparent tokenization model for computing power, it could disrupt the cloud computing market for AI.