TL;DR
- Is exiting the cloud really the best way to save on infrastructure costs?
- How much are AWS, Google, and Microsoft spending on gen AI?
- Top announcements from re:Invent, KubeCon, and Ignite
Keep reading to find out what happened in the cloud over the last month.
_____
Story of the Month
Is A Cloud Exit The Ultimate Way To Slash Cloud Costs?
X shifted its workloads to on-prem and reduced the monthly cloud costs by 60%, DHH announced.
X used to spend nearly $100 million/year on AWS, so does this mean that the company is on track to save $60 million annually?
Actually, this raises more questions than provides answers.
- Is this really a cloud exit or just a shift to a hybrid environment?
- How much of the $60 million in cost savings has now become CapEx instead of OpeEx?
- What about the expertise required of DevOps and IT teams to make the most of on-prem?
Making this move might be viable for a company like X, but dynamic companies that expect rapid growth better stick to the cloud. That’s because predicting their growth rate is next to impossible.
The potential lag time that comes with ordering, setting up, and maintaining servers just isn’t sustainable. Purchasing too much capacity at the beginning isn’t a good idea either.
Staying in the cloud makes sense, especially since you can use automation as a core part of your cost optimization initiatives and get much better results than shifting to on-prem. Here’s a case in point.
_____
The Cloud in Numbers
The Impact Of Gen AI On The Cloud
$2 billion – this is how much Google will invest in Anthropic. Amazon is already investing $4 billion in it to catch up with the AI game.
$116 billion – Bank of America analysts expect the cloud-related CapEx of Amazon, Alphabet and Microsoft to reach this number next year, growing by 22%.
$42 billion – this is how much the three major cloud providers spent in Q3 2023 on building capacity to serve gen AI, 20% more than in the same period in 2021.
40% – that’s the chunk of respondents who mentioned “reducing time spent on managing cloud services” as one of the top benefits of their cloud cost optimization initiatives (for 70%, it was all about reducing cloud costs).
____
Top announcements from AWS re:Invent, KubeCon, and Microsoft Ignite
- Amazon SageMaker Clarify helps to evaluate and select foundation models
- Amazon Q, a new generative AI-powered assistant
- AWS introduced a Cost Optimization Hub that centralizes cost recommendations
- New Amazon S3 Express One Zone high-performance storage class
- Four new additions to Microsoft’s Copilot gen AI service offerings suite
- Microsoft has two new internally-made chips: Azure Maia 100 AI Accelerator and the Azure Cobalt 100 CPU
- Microsoft Defender XDR and Sentinel are now equipped with Security Copilot
- CAST AI announced its $35-million Series B round round at KubeCon. 🚀
____
The Business of Cloud
Chip Stories Continue, But Microsoft Enters The Game
Overwhelmed with demand for GPUs to train and run gen AI models, Nvidia is using its power to build long-term moats around software ecosystems, AI platforms, and a cloud platform with Lenovo and ARM chips. Link, Link
Microsoft said it will start using custom-designed chips for AI in its data centers. Link
AWS launched a new service called Amazon EC2 Capacity Blocks for ML to democratize ML by helping teams get GPU-powered instances for training and inference. Link
_____
Security & Outages
Differences in Hyperscaler Outage Transparency
The Pragmatic Engineer compared the responses from AWS, Azure, and Google to regional outages that happened between April and July 2023. Link
_____
This series explores the most attention-grabbing cloud technology news, bringing you up to speed with the latest releases, acquisitions, research, and hidden gems in cloud computing – the stuff actually worth reading.
Get curated cloud industry news straight to your inbox
A monthly roundup of the most notable things going on in the cloud industry. Subscribe below and stay up to date.