Eight Sleep's cloud-only Pod3 became a $2,000 'sauna' during the AWS outage

Eight Sleep sells AI-powered mattress covers that track heart rate and sleep stages and adjust temperature through water-cooled coils. The Pod3's real-time adjustments, data syncing and preset routines all ran through the company's backend servers, which sat on Amazon Web Services — there was no offline mode.

On October 20, 2025 an AWS disruption in its US-EAST-1 region rippled across the internet from around 3 a.m. ET, taking down Snapchat, Roblox and Fortnite alongside it. Eight Sleep's support channels filled with owners whose app froze on loading screens, whose Hub touch panels went unresponsive, and whose beds were stranded in whatever thermal setting was last set. Reddit threads turned into choruses of 'my bed is bricked', with some Pods unable to reboot without cloud clearance.

The emblematic case was tech enthusiast Alex Browne, who had programmed his Pod3 to pre-emptively warm by 9 degrees Fahrenheit before bedtime. When AWS went dark the system locked into that preset with the cooling override disabled, and he spent the night tweeting updates: 'Backend outage means I'm sleeping in a sauna.' The post went viral.

Co-founder and CEO Matteo Franceschetti apologized on X, said the team was restoring features as AWS recovered, and committed to 'outage-proofing' the Pod experience going forward.

Pod3 covers cost about $2,000; during the outage users lost temperature control, sleep tracking and all automations, some beds frozen until cloud clearance.

Every feature that justified the premium price — temperature control, tracking, routines — depended on Eight Sleep's cloud servers rather than the device.

The physical controls were designed as app backups, not standalone controls, so they glitched or failed when the cloud did.

No offline fallback existed even for core functions, despite a device price of roughly $2,000.

A connected product needs a way to keep its basic promise offline; otherwise you have outsourced the product's core function to someone else's uptime.

Features came back as AWS services recovered, and Franceschetti promised the team would work around the clock to make sure the outage never disrupted sleep again — a pledge many users read as a commitment to offline-capable systems. The episode became the standard citation for the risks of hard cloud dependency in connected hardware.