---
title: "Acrab’s $130M Series B: On-Device Agentic Compute Leaves the Cloud"
description: "Vertex Ventures SEA & India and Vertex Growth backed Singapore’s Acrab in a $130M Series B to commercialize GΞLIX 1 Agent Box — silicon + edge models + orchestration for agents."
date: 2026-08-06T00:00:00.000Z
tags: ["2026-vc-news", "startup-funding", "venture-capital", "artificial-intelligence", "ai-infrastructure", "asia-vc"]
source: https://venturecapitaltracker.com/2026-acrab-130m-series-b-agentic-edge-compute
---

# Acrab’s $130M Series B: On-Device Agentic Compute Leaves the Cloud

> Vertex Ventures SEA & India and Vertex Growth backed Singapore’s Acrab in a $130M Series B to commercialize GΞLIX 1 Agent Box — silicon + edge models + orchestration for agents.

**Acrab** raised a **$130 million Series B** on **August 6, 2026**, with continued support from **Vertex Ventures SEA & India** and **Vertex Growth**, plus European and Southeast Asian institutions. The Singapore company is commercializing **on-device agentic compute** — silicon, edge models, and orchestration — not another cloud API wrapper.

## Key facts

| Field | Detail |
| --- | --- |
| Company | Acrab (Singapore) |
| Round | $130M Series B |
| Date | August 6, 2026 |
| Lead / key backers | Vertex Ventures SEA & India, Vertex Growth |
| Founded | 2024 (Tian Qi) |
| Cumulative funding | >$350M at/after stealth emergence (company); some press cite ~$480M post-round |
| Product | GΞLIX 1 SoC + Agent Box platform |
| Near-term claim | Industrial deployment paths; expected revenue within 2026 (company) |

## Who uses the product — and for what job

**Users (target):** device and industrial OEMs, edge/robotics integrators, and product teams that need agents to **act locally** — privacy, latency, offline, or bandwidth constraints.

**Job:** run personalized, real-time agent behavior at the edge without round-tripping every token to a hyperscaler GPU cluster.

Acrab’s stack claim is vertical: **custom silicon + edge models + orchestration software**. Agent Box is the commercialization package after the GΞLIX 1 unveil.

## Why now

- Cloud agent demos hit cost, latency, and data-residency walls in physical environments.
- “Agentic” moved from chat UIs to machines that must sense and act in milliseconds.
- Asia’s hardware + software builders can ship full stacks faster than cloud-only peers for devices.
- Investors want AI infra beyond data-center GPUs — devices and industrial systems are the next billable surface.

## Why Vertex — portfolio fit

Vertex’s SEA/India early-stage franchise plus Vertex Growth is a classic **region-native double-down**: finance the stealth hardware/software build, then fund commercialization when industrial paths appear.

**Likely founder rationale:** keep the investor that already underwrote multi-hundred-million stealth capital and understands SEA hardware go-to-market — rather than reset the board with a brand-name U.S. infra fund that has never diligenced the silicon.

| Dimension | Fit |
| --- | --- |
| Geography | Singapore HQ, SEA manufacturing adjacency |
| Stage | Growth commercialization after stealth |
| Thesis | Agentic compute at the edge, not only in the cloud |
| Risk | Hardware + model + GTM execution in one vehicle |

We do not currently list Vertex Ventures SEA as a `/fund/` entity — names are press-sourced.

## Competitive map

| Player | Lane |
| --- | --- |
| NVIDIA Jetson / edge GPU stacks | Incumbent programmable edge AI |
| Qualcomm / mobile NPUs | On-device inference at phone/IoT scale |
| Cloud agent platforms | Opposite deployment polarity |
| Other custom AI ASIC startups | Silicon without full agent software stack |

## Market signal

**$130M on top of $350M+ cumulative** is a statement that Vertex believes **edge agent hardware** can clear industrial revenue in 2026. Absence of a disclosed valuation means you should not invent one for memos.

## When not to model this as a SaaS deal

- Wrong if you underwrite like a pure software ARR story — silicon cycles dominate.
- Wrong if “expected revenue this year” is treated as booked ARR.
- Wrong if cloud-only agent startups copy the pitch without a device channel.

## Practical takeaway

- **Founders (edge AI):** Pair silicon claims with named industrial design-ins; capital markets punish vapor SoCs.
- **Investors:** Separate cumulative stealth capital from commercial Series B milestones (deployments, yield, revenue).
- **Operators:** Ask what runs offline, what still phones home, and who owns model updates on the device.

### Sources

1. PR Newswire (Aug 6, 2026): https://www.prnewswire.com/news-releases/acrab-raises-us130-million-series-b-advancing-agentic-ai-compute-platform-commercialization-302844535.html
2. DealStreetAsia: https://www.dealstreetasia.com/stories/ai-startup-acrab-raises-130m-in-vertex-backed-series-b-funding-491373
3. Related: [/2026-august-9-investment-news-roundup-agents-voice-health](/2026-august-9-investment-news-roundup-agents-voice-health)