Brainwave | Private AI hardware for teams

Solutions

Built for the way
your team works

One Brainwave serves everyone on the network, so you don't need a high-spec machine or an AI subscription for every person.

01 / 03

Capabilities

Everything runs
on Brainwave

Nine capabilities come built in. Nothing to install, no plugins to manage, and no usage caps.

brainwave.local

Summarise last week's sales from our local database.

847 transactions totalling $124,390, up 12% on the prior week. The largest gain came from repeat customers in the northeast region.

Answered on-device · no data left the network

Turn that into a one-page PDF for Monday's meeting.

Ask anything…↑

01

AI Chat

A full chat assistant running local language models, for everyone on the network.

02

Work

Shared workspaces where your team and the AI work on the same documents in real time.

03

CLI

Script, automate and query Brainwave from the terminal, or point your own tools at it.

04

Local database

Connect your databases, documents and files. The AI reasons over them without sending them anywhere.

05

Automation workflows

Scheduled and triggered workflows that handle the repetitive work in the background.

06

Image generation

Text-to-image with local diffusion models. No credits and no limits.

07

Video generation

Short-form video from text or images, rendered on Brainwave instead of in the cloud.

08

Transcription

Accurate speech-to-text for meetings and calls. Searchable, exportable, private.

09

Real-time translation

Live speech and text translation for international teams and client meetings.

Apps

One Brainwave.
Every screen.

Desktop, web and mobile apps for everyone on your network. Every request is answered by the Brainwave in your office.

Brainwave desktop app: Real-time translation

Brainwave iPhone app: Real-time translation

Start anything from one prompt, and see your Brainwave's status at a glance.

Design

Engineered to live
in your space

An aluminum enclosure with a ribbed ventilation architecture, built to sit on a desk, a shelf or a meeting-room table, and to be seen.

Top detail

Rear I/O

Power and ventilation

360° view

Machined detail

Enclosure: Aerospace-grade aluminum
Cooling: Passive + active hybrid
Finishes: Silver, Copper, Gold
Connectivity: 10 Gb Ethernet, Wi-Fi 6

Models

Three models.
One for every team.

Each finish carries its own tier of compute, from a compact AI server for a small team to workstation-class power for a whole department.

Silver

Essential

Polished aluminum. A compact AI server for individuals and small teams.

Graphics: NVIDIA RTX Spark N1X, Blackwell GPU with 6,144 CUDA cores
Graphics memory: 128 GB LPDDR5X unified
Memory bandwidth: 300 GB/s
AI performance: 1 PFLOP FP4 (≈1,000 AI TOPS)
Processor: NVIDIA Grace, 20 cores (Arm)
System memory: Shared with graphics (128 GB unified)
Storage: 2 TB NVMe SSD
Networking: 10 Gb Ethernet, Wi-Fi 6
Largest model: Up to 120B parameters
Best for: Individuals and teams up to 10
Warranty: 2 years
Estimated delivery: February 2027

Reserve Silver

Copper

Pro

Rich copper. Dedicated workstation graphics for fast responses across a busy team.

Graphics: NVIDIA RTX PRO 4000 Blackwell
Graphics memory: 24 GB GDDR7 ECC
Memory bandwidth: 672 GB/s
AI performance: 1,290 AI TOPS
Processor: AMD Ryzen 7 9700X, 8 cores / 16 threads
System memory: 64 GB DDR5
Storage: 2 TB NVMe SSD
Networking: 10 Gb Ethernet, Wi-Fi 6
Largest model: Up to 32B parameters
Best for: Teams up to 25
Warranty: 2 years
Estimated delivery: November 2026

Sold out

Gold

Max

Warm champagne. Workstation-class power to serve a whole department at once.

Graphics: NVIDIA RTX PRO 6000 Blackwell
Graphics memory: 96 GB GDDR7 ECC, 512-bit
Memory bandwidth: 1,792 GB/s
AI performance: ≈4,000 AI TOPS
Processor: AMD Ryzen 9 9950X, 16 cores / 32 threads
System memory: 128 GB DDR5
Storage: 4 TB NVMe SSD
Networking: 10 Gb Ethernet, Wi-Fi 6
Largest model: Up to 120B parameters, at full GPU speed
Best for: Departments of 50+
Warranty: 3 years
Estimated delivery: January 2027

Reserve Gold

AI performance is NVIDIA's peak FP4 rating with sparsity. Largest model assumes 4-bit quantization. Every model includes the full Brainwave software stack, multi-user accounts and the admin dashboard, with no subscription.

Setup

Up and running
in minutes

No IT department required. Brainwave is designed to be unboxed, connected and shared by anyone.

01

Plug in

Connect power and your router. Brainwave boots in under 30 seconds.

02

Connect

It appears on your network automatically. Open the app on any device and connect in one tap.

03

Invite your team

Add teammates, set permissions and connect data sources from the admin dashboard.

04

Start working

Every capability is available immediately, from every device on the network.

Why Local

Private by design.
Powerful by default.

Enterprise-grade AI for organizations of any size, without trading away privacy, speed or cost.

0

Round trips to the cloud. Every request is answered on your network.

∞

No usage limits or rate throttling.

$0

Ongoing cloud fees after purchase.

24/7

Always on, always available to your team.

Private AI
for your whole team

Built for the way
your team works