IBM plans $5B Internet of Things push, lures The Weather Channel away from AWS
John Cook / GeekWire :
Context & Ripple Effects
In March 2015 IBM committed $5 billion to its Internet of Things push, and its first marquee win was pulling The Weather Channel's workloads off AWS onto IBM Cloud. Within months the customer relationship turned into ownership: IBM agreed to buy [[a:835213|The Weather Co.'s digital and data assets — not its TV channel — in a deal sources valued above $2B]], closing it in January 2016 and installing Weather CEO David Kenny atop the Watson platform [[a:864100]].
The arc since then reads as a full capital cycle: IBM extended the IoT bet by embedding Watson into Cisco's edge routers [[a:870206]], but by 2023 it was shopping the weather unit at a reported $1B-plus ask — well under what it reportedly paid — before agreeing to sell Weather.com and the business-oriented services to Francisco Partners [[a:843359]]. This 2015 announcement is where that cycle began.
First-order effects
- The Weather Channel immediately migrates its business from AWS to IBM Cloud, handing IBM a high-profile tenant and a continuous stream of real-world sensor and forecast data to feed its new IoT division.
- AWS loses one of its most recognizable consumer-data workloads, while IBM gains a reference customer it can point to when selling against Amazon in enterprise IoT pitches.
Second-order effects
- The win converts into an acquisition: rather than remaining a cloud customer, Weather Co.'s digital assets are absorbed into IBM, turning the anchor tenant into proprietary training and analytics fuel for Watson.
- IBM pairs the platform with Cisco, pushing Watson analytics into edge routers and forcing rivals to respond with their own hardware-plus-AI bundles rather than cloud-only offerings.
Third-order effects
- The pattern — court a flagship customer, acquire its data assets to seed a platform, then divest when the thesis fades — ended with IBM exiting at a reported loss versus its ~$2B+ outlay, suggesting conglomerate AI platforms built through data acquisitions rarely hold their entry valuation.
- If the pattern holds across the industry, vertically integrated IoT/AI stacks assembled during platform-building phases get unbundled back to specialist private-equity owners, with the original acquirer retaining only the core software layer.
The trend: Big-platform AI strategies built by acquiring data-rich customers tend to run a full capital cycle — premium entry price, integration into the stack, then divestiture to specialists once the synergies underdeliver.