A breakdown of Google Cloud's rivalry with AWS and Azure as analysts expect $26B in 2022 GCP revenue and IDC predicts ~30% YoY cloud market growth until 2025
Nico Grant / Bloomberg : Tweets: @nicoagrant Tweets: Nico Grant / @nicoagrant : New details here about Google's Project Drawbridge and Project Moat, efforts to bolster Google Cloud's reliability after high-profile outages. It turns out the global network that keeps Search running isn't always ideal for the cloud. In the next @BW issue https://www.bloomberg.com/...
Context & Ripple Effects
Google Cloud has spent years trying to close the gap with AWS and Azure — the Diane Greene era was defined by that catch-up struggle and tension inside Google over strategy. The new Bloomberg reporting adds a technical wrinkle: the global network built for Search turns out to be a liability for cloud workloads, which is why Google is now running Project Drawbridge and Project Moat to shore up reliability after high-profile outages.
The stakes are set by the numbers in the same coverage: analysts expect $26B in 2022 GCP revenue while IDC projects roughly 30% annual cloud-market growth through 2025 — a market expanding fast enough that reliability missteps cost share, not just customers.
First-order effects
- Google Cloud commits engineering resources to Projects Drawbridge and Moat, effectively building cloud-dedicated network capacity rather than riding the Search backbone — directly addressing the outage-driven trust deficit with enterprise buyers.
- AWS and Azure face a rival whose $26B run-rate keeps it in the conversation even from third place, keeping pricing and enterprise-deal competition live across all three.
Second-order effects
- In a market IDC expects to grow ~30% annually through 2025, reliability becomes a competitive weapon: AWS and Azure can be expected to press their own uptime records in enterprise sales where Google's outages give procurement teams a reason to stay put.
- Google's later strength in AI workloads — winning business from labs like OpenAI and Thinking Machines per related coverage — depends on this reliability foundation holding, since training runs amplify the cost of any infrastructure failure.
Third-order effects
- If Drawbridge and Moat succeed, they formalize a structural split inside hyperscalers between consumer-scale networks and cloud-grade infrastructure — an admission that scale built for one workload doesn't transfer free to another.
- Sustained 30% market growth through 2025 keeps the race a land-grab rather than a zero-sum share fight, meaning all three hyperscalers can grow while the real contest shifts to which workloads — AI first among them — each platform locks in.
The trend: Hyperscalers are learning that consumer-internet infrastructure does not automatically serve cloud customers, pushing Google, AWS, and Azure toward purpose-built cloud networks as AI workloads raise the reliability bar.