The largest edge network was hiding in plain sight.

Every Sensify cooler now carries an E1 compute node.

Talk to us →
Nodes
312,480
Aggregate compute
31.2 EOPS
Countries
64

Every cooler you walk past is now a computer.

Compute where data is born.

Latency
<10ms p50
Density
300k+sites
Locality
0pixels out
Cloud region
~80 ms
Metro edge
~25 ms
Sensify E1
~4 ms
Round-trip inference latency, typical urban deployment. Illustrative.

One network.
Everywhere.

Nodes
312,480
Countries
64
Compute
31.2 EOPS
Drag to rotate
Sensify Edge E1: the liquid-cooled GPU bay on top of the cooler

E1. A datacenter rack, cooler-sized.

Built into every Sensify cooler. Liquid-cooled, sealed, on the circuit already there.

Thermal01
Closed-loop liquid cooling
Compute02
3 × full-length GPUs, water-blocked
Host & storage03
Server-class CPU · hot-swap drive cage
Network04
5G / LTE · Wi-Fi 6 · local mesh
Camera I/O05
Native shelf + cooler camera inputs
Security & power06
TPM 2.0 · secure boot · dedicated PSU

Three layers. Pick your depth.

Compute

Raw E1 capacity by node, region or hour. Bring a container, get a GPU next to the data.

Talk to us →

Platform

Orchestrate distributed workloads across the fleet. One deploy, every store in a region.

Talk to us →
Store 0412 · Shelf 3Live
Out of stockCola 500 ml · 3 facings
Planogram compliance91%

Solutions

Managed services built by Sensify: shelf vision, out-of-stock detection, in-store media.

Talk to us →

One city.
One computer.

What runs at the last meter.

OOS · 0.94
FACING · 12

Shelf vision & out-of-stock

Detect gaps, planogram drift and pricing errors the moment they happen.

Local LLM inference

Small language models served within a few hundred meters of every user.

Federated learning

Train across thousands of stores without moving a single raw record.

In-store media

Screens and offers rendered locally, measured locally, in real time.

Urban sensing

Foot traffic, air quality and mobility signals from a city-wide mesh.

Demand forecasting

Per-store predictions computed where the sales actually happen.

Pixels stay in the store.

Cameras feed E1 directly. Inference runs on-device. What leaves is a count, an event or an embedding. Never a face.

In-store CameraRaw frames E1 InferenceOn-device CloudMetadata only

One command. Every store in a city.

Containers in. GPUs out. Target by region, retailer or tag, and roll out like any cloud.

Request early API access →
# deploy a vision model to every node in São Paulo
$ edge deploy ./shelf-vision \
    --region br-sp \
    --accelerator gpu \
    --replicas all \
    --rollout canary:5%

→ resolving nodes…
✓ 18,204 nodes matched
✓ canary healthy · p50 4 ms
✓ rolled out to 100%
from sensify_edge import Fleet

fleet = Fleet(token="…")

job = fleet.deploy(
    image="ghcr.io/acme/shelf-vision:1.4",
    target={"region": "br-sp"},
    accelerator="gpu",
    rollout="canary:5%",
)

for event in job.stream():
    print(event.node, event.latency_ms)
POST /v1/deployments
Authorization: Bearer …

{
  "image": "ghcr.io/acme/shelf-vision:1.4",
  "target": { "region": "br-sp" },
  "accelerator": "gpu",
  "rollout": "canary:5%"
}

201 Created · 18,204 nodes matched

Same cooler. Same price. Now it's infrastructure.

Edge subsidizes the Sensify controller. You pay what you paid before. Your store joins a global network.

Become a host →
Store owner next to a Sensify cooler with an E1 node

Every cooler, a node.

Today312k nodes, 64 countries
NextEvery Sensify cooler
Then1M+ nodes
HorizonThe last meter, everywhere

Build on the last meter.

Onboarding the first builders, brands and hosts.

Thanks. You're on the list — we'll be in touch.