How much did Acrab raise?

Acrab raised $130 million in Series B funding announced August 6, 2026. Cumulative funding is reported above $350 million since emerging from stealth (some outlets cite ~$480M including the new round).

Who invested in Acrab’s Series B?

Existing investors Vertex Ventures Southeast Asia & India and Vertex Growth continued as lead backers, with additional institutional investors from Europe and Southeast Asia.

What does Acrab build?

A full-stack agentic AI compute platform combining purpose-built silicon (GΞLIX 1), edge AI models, and software orchestration — packaged as Agent Box for real-time on-device agents.

Industrial and consumer device OEMs, robotics/edge deployers, and enterprises that need agents to run locally with low latency and less cloud dependence.

Why raise from Vertex again?

Vertex’s SEA/India and Growth funds already financed the stealth build. A commercialization Series B from the same platform reduces syndicate risk while Acrab proves industrial deployments and 2026 revenue.

Is a valuation disclosed?

No public valuation was disclosed in the August 2026 Series B materials reviewed for this note.

Acrab $130M Series B — Agentic Edge Compute

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

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
DealStreetAsia: https://www.dealstreetasia.com/stories/ai-startup-acrab-raises-130m-in-vertex-backed-series-b-funding-491373
Related: /2026-august-9-investment-news-roundup-agents-voice-health

Venture Capital Tracker

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