Welcome to episode 294 of The Cloud Pod – where the forecast is always cloudy!Ilya Boy, do we have a news packed week for you! Sutskever raised $30B without a product, Mira Murati launched her own AI lab, and Claude 3.7 now thinks before it speaks. Meanwhile, Microsoft casually invented new matter for quantum computing, Google built an AI scientist, and AWS killed Chime (RIP). At this rate, AI is either going to save the world or speedrun becoming Ultron. Let’s all find out together – today on The Cloud Pod! 

Titles we almost went with this week:

• ☠️Ding – Chime is Dead
• 🫙Does your container really need 192 cores
• 💻Quantum is the new AI
• 🤖AI is now IN the robots

A big thanks to this week’s sponsor:

We’re sponsorless! Want to get your brand, company, or service in front of a very enthusiastic group of cloud news seekers? You’ve come to the right place! Send us an email or hit us up on our slack channel for more info. 

AI Is Going Great – Or How ML Makes All It’s Money 

02:41 Ilya Sutskever’s Startup in Talks to Raise Financing at $30 Billion Valuation

• It’s been a minute since we talked about former OpenAI executives and what they’re up to. 
• Let’s start with Ilya Sutskever and Mira Murati, post Open AI career
• The Information reports that Ilya Suskevers’ startup “Safe Superintelligence” is in talks to raise $1Billion in a round that would value the startup at $30 Billion.  
• The company has yet to release a product, but based on the name we can guess what they’re working on…

03:22 📢 Ryan – “It’s so nuts to me that they can raise that much without – really just an idea. Doesn’t have to have any proof or POC…”

07:07 Murati Joins Crowded AI Startup Sector

• Mira Murati confirmed one of the worst kept secrets in AI, by revealing her lab Thinking Machine Labs. 
• Murati has lured away two thirds of her team from OpenAI. 
• We’ll be waiting to see how the funding goes for this one. 

08:02 Claude 3.7 Sonnet and Claude Code

• Anthropic is releasing their latest model Claude 3.7 Sonnet, their most intelligent model to date and the first hybrid reasoning model on the market.  
• Claude 3.7 sonnet can produce near instant responses or extended, step by step thinning that is made visible to the user.  
• API users also have fine grained control over how long the model can think for. 
• Claude 3.7 shows particularly strong improvements in coding and front-end web development.  
• In addition to the new model they have introduced a command line tool for Agentic Coding, Claude Code.  
• Claude code is available as a limited research preview and enables developers to delegate substantial engineering tasks directly from the terminal (Justin really wants a native VS code integration… come on Anthropic!)
• The extended thinking model is not available in the free tier, but all other paid plans are covered as well as through our various cloud providers. 
• Want to join the preview? You can do that here. 

12:44 📢 Justin – “AI is great. I can see how it makes good coders even better, and bad coders worse, and your ability to be a debugger is gonna be the make or break for you in the AI coding world.”

AWS

14:58 Update on Support for Amazon Chime 

• Amazon has decided to end support for their Amazon Chime service, including business calling features, effective February 20th, 2026. 
• Amazon Chime will no longer accept new customers starting February 19th, 2025. 
• You can continue to use it as an existing customer for meetings through February 20, 2026, and you can delete your data prior to that day. 
• For those of you using the Amazon Chime SDK, this service will not change (it powers Slack Huddles.)
• Amazon provides you with a few options to replace Chime including their own AWS Wikr service, or from AWS partners such as Zoom, Webex and Slack. 

16:38 📢 Matthew – “I was surprised at how short of a timeline this was, because I feel like code command and some of the other ones are multi-year, and maybe that’s just memory. But one year, if you’re fully integrated into the solution, doesn’t feel like a long time to migrate as a business. Or no one actually uses it, so who cares? One of the two.”

19:07 Amazon ECS increases the CPU limit for ECS tasks to 192 vCPUs

• In today’s “Are you sure containers are the right solution,” ECS now supports CPU limits of 192 vCPU’s for ECS tasks deployed on EC2 instances, an increase from the previous 10 vCPU limit.  
• This enhancement allows customers to more effectively manage resource allocation on larger Amazon EC2 instances. 

21:09 Anthropic’s Claude 3.7 Sonnet hybrid reasoning model is now available in Amazon Bedrock

• Sonnet available in Bedrock and Q Developer.  
• Finally. 

22:50 AWS Network Firewall introduces automated domain lists and insights

• One less thing we have to use Athena for… is a victory in our book! 
• AWS Network Firewall now offers automated domain lists and insights, a feature that enhances visibility into network traffic and simplifies rule configurations.
• The capability analyzes HTTP and HTTPS traffic from the last 30 days, and provides insights into the frequency of access to domains, allowing a quick rule creation based on observed network traffic patterns. 

23:10 📢 Ryan – “It’s funny because I, when they rolled out this, this feature or the network firewall together, I’ve become real spoiled. And so like, when it didn’t have this, was like, how am I supposed to use this? I gotta compile all my traffic to figure out what’s going on. Like, boo. And so, yeah, this is, this is great because compiling these data sets and running your queries is a chore. And typically that’s all you want, right? You just want to be able to very quickly sort of say this is what’s coming in and answer a question and move on.”

GCP

26:38 Introducing Cloud DNS public IP health checks, for more resilient multicloud deployments

• Google is announcing the General Availability of Cloud DNS routing policies with public health IP checking, which provides the automated, health-aware traffic management that you need to build resilient applications, no matter where your workloads reside. 
• Running on multiple cloud providers can often lead to fragmented traffic management strategies. Cloud DNS now lets you intelligently route traffic across multiple cloud providers based on application health from a single Interface.  Cloud DNS supports a variety of routing policies, including weighted round robin, geolocation, and failover, giving you the flexibility to tailor your traffic management strategy to your specific needs. 

27:03 📢 Ryan – “I mean, so maybe you can take your Kubernetes workload and actually spread it across multiple clouds and serve from all clouds with a solution like this. it’s always that sort of edge case where it’s sort of the rubber meets the road and you run into these weird things trying to serve from multi-cloud. But this is a big step towards that. I’m sure there’s other edge cases that I’m not thinking about. I know there’s a ton of operability concerns, but this is kind of neat.”

30:13 Announcing quantum-safe digital signatures in Cloud KMS 

• Google is releasing quantum-safe digital signatures (FIPS 204/FIPS 205 compliant) in KMS for software-based keys, available in preview. 
• They are also sharing their high-level view in their post quantum strategy for google cloud encryption products, including for Cloud KMS and Cloud HSM. 
• Their goal is to ensure that Google Cloud KMS is quantum safe
  • Offering software and hardware support for standardized quantum-safe algorithms
  • Supporting migration paths for existing keys, protocols and customer workloads to adopt PQC.
  • Quantum-proofing Google’s underlying core infrastructure
  • Analyzing the security and performance of PQC algorithms and implementations. 
  • And contributing technical comments to PQC advocacy efforts in standard bodies and government organizations. 

33:04 📢 Justin – “They implemented hybrid plus quantum key exchange that provides traditional and quantum resistant algorithms, and they provided protection against both current threats and potential future quantum computer-based attacks. And then it goes on to talk about the implementation use of Kyber. And I do remember us talking about Kyber because I think we talked, we talked about Kyber crystals at one point. Yeah. So we did talk about this at one point. So yes, you’re good already. So yes, Google is coming in either, maybe behind Azure. I don’t know. I don’t know if they have anything.”

34:47 Introducing A4X VMs powered by NVIDIA GB200 — now in preview

• Google is announcing the preview of A4X VMs, powered by NVIDIA GB200 NVL72, a system consisting of 72 NVIDIA Blackwell GPUs and 36 arm-based NVIDIA Grace CPus connected via fifth generation NVIDIA NVLink.  
• With this integrated system, A4X VMs directly address the significant compute and memory demands of reasoning models that use chain-of-thought, unlocking new levels of AI performance and accuracy.
• Google Cloud is the first and only provider today to offer both the A4 Vm powered by NVIDIA B200 and A4x VMs powered by NVIDIA GB200 NVL72.
• Google says to help you choose the best one for your workload:
  • A4X VMs (powered by NVIDIA GB200 NVL72 GPUs): Purpose-built for training and serving the most demanding, extra large-scale AI workloads, particularly those involving reasoning models, large language models with long context windows, and scenarios that require massive concurrency. This is enabled by the unified memory across a large GPU domain.
  • A4 VMs (powered by NVIDIA B200 GPUs): A4 provides excellent performance and versatility for diverse AI model architectures and workloads, including training, fine-tuning, and serving. A4 offers easy portability from prior generations of Cloud GPUs and optimized performance benefits for varying scaled training jobs.

36:07 📢 Justin – “So basically, if you need big, expensive hardware, use the A4X. And if you want to do some inference and basic things, you have a model you’re already happy with, the A4 VM is probably the right choice for you.”

37:40 We’re launching a new AI system for scientists

• Google has launched an AI co-scientist, a new AI system built on Gemini 2.0 designed to aid scientists in creating novel hypotheses and research plans.  
• Researchers can specify a research goal for example, to better understand the spread of disease-causing microbes using natural language and the AI co-scientist will propose testable hypotheses, along with a summary of relevant published literature and a possible experimental approach.  

38:02 📢 Ryan – “That’s wild – it’s such a specific use case!” 

39:59 Announcing Claude 3.7 Sonnet, Anthropic’s first hybrid reasoning model, is available on Vertex AI

• Claude 3.7 is available on Vertex AI
• Vertex apparently also supports Claude Code.  
• Shocking. 

Azure

41:33 Microsoft quantum breakthrough promises to usher in the next era of computing in ‘years, not decades’ 

• Microsoft has announced they have created a new type of matter, growing up in sciences, you would learn that there are three main types of Matter including Solids, Liquids and Gas. But now Microsoft has turned this on its head. 
• They have created an entirely new state of matter, unlocked by a new class of materials, topoconductors that enable the fundamental leap in computing.
• All of this powers Majorana 1, the first quantum processing until built on topological core. 
• Satya believes this breakthrough will allow them to create a truly meaningful quantum computer not in decades but in years. 
• The qubits created with topoconductors are faster, more reliable and smaller. They are 1/100th of a millimeter, meaning we now have a clear path to a million-qubit processor. 
• A chip the size of the palm of your hand yet is capable of solving problems that even all the computers on earth today could not. 

42:37 📢 Ryan – “Wow. I mean, that last bullet point is what my head explodes. Like I know I don’t understand quantum computers and I, you know, like from any kind of way. Now, you know, they’re introducing new states of matter, like in order to power some of those things, it’s gonna, it just feels like tomorrow world is gonna be completely unrecognizable.”

45:28 Microsoft’s Majorana 1 chip carves new path for quantum computing

• Microsoft says they took a step back, and said OK if we invent the transistor for the quantum age. What properties does it need to have? And that’s apparently how they got there. 
• Being able to fit a million qubits in the palm of the hand, unlocks the path to meet the threshold for quantum computers to deliver transformative, real-world solutions such as breaking down microplastics into harmless byproducts or inventing self-healing material for construction, manufacturing and healthcare. 
• All of the current computer’s operating together can’t do what a one-million-qubit quantum computer will be able to do. 
• The first topological core powering the Morjana 1 is reliable by design, incorporating error resistance at the hardware level making it more stable. 
• Commercially important applications will require trillions of operations on a million qubits, which would be prohibitive with current approaches that rely on fine-tuned analog control of each qubit.  
• The new chip allows them to be controlled digitally, redefining and vastly simplifying how quantum computing works.
• Microsoft is now one of two companies to be invited to the final phase of DARPA’s underexplored systems for utility quantum computing program.  

47:52 Microsoft’s new AI agent can control software and robots

• Microsoft research unveiled Magma, an integrated AI foundational model that combines visual and language processing to control software interfaces and robotic systems. 
• If the results hold up outside of MS, it could be a meaningful step forward for an all purpose multi-model AI that can operate interactively in both real and digital spaces. 
• MS claims its the first multi-modal AI model that can not only process but also actively act on the data from navigating user interfaces to manipulating physical objects.  

49:30 📢 Justin – “Well, good, I look forward to our future robot overlords.”

49:36 Introducing Azure AI Foundry Labs: A hub for the latest AI research and experiments at Microsoft

• Azure is launching Azure AI Foundry Labs, a hub for developers, startups and enterprises to explore groundbreaking innovations from research to Microsoft. (until Visual Studio kills it…. I’m watching you VS studio)
• Microsoft’s newest AI breakthrough Muse, is a first of its kind world and human action model (WHAM), available today in Azure AI Foundry, is the latest example of bringing cutting-edge research innovation to their AI platform for customers to use. 
• With Azure AI foundry labs they are excited to unveil new assets for their latest research driven projects that empower developers to explore, engage and experiment.  Projects across models and agentic frameworks include:
  • Aurora – A large-scale atmosphere model providing high-resolution weather forecasts and air pollution predictions, outperforming traditional tools. 
  • ExACT: an open source project enabling agents to learn from past interactions and improve search efficiency dynamically
  • Magnetic-One: a multi-agent system involving complex problems by orchestrating multiple agents, built on the autogen framework.
  • Mattersim: a deep learning model for atomistic simulations, predicting material properties with high precision
  • OmniParserver v2: a vision-based module converting UI screenshots into structured elements, enhancing agents’ action generation.
  • TamGen: a generative AI model for drug decision, suing a GPT-like chemical language model for target-aware molecule generations and refinement
• Microsoft points out that the speed of innovation is crucial, and points to the slow adoption of GPS and the decade it took from military applications to consumer use.  But AI innovations are moving much faster than that.
• The pace of AI advancement has accelerated dramatically. 

50:55 📢 Ryan – “I mean, I can’t agree more with the speed of innovations blinding. You know, we started this podcast to keep up with cloud news as the hyperscalers got to a certain scale where they were announcing enough stuff that we couldn’t keep up to date. Now I feel even with this, it’s, I really struggle to, you know, understand half of these use cases and how it’s applied and the whole thing. Like it’s crazy to me how fast things are moving.”

51:46 A new level unlocked

• Microsoft is also releasing Muse, a first of its kind generative AI model that they are applying to gaming. It’s a huge step forward in gameplay ideation. 
• Muse, just from observing human gameplay, has developed a deep understanding of the environment, including its dynamics and how it evolves over time in response to actions.  This unlocks the ability to rapidly iterate, remix and create in video games so developers can eventually create immersive environments and unleash their full creativity. 

Aftershow

1:03:02 Amazon Paid $1 Billion for Control of Bond Franchise 

• Amazon paid around $1billion to secure creative control of the James Bond franchise, according to a person familiar with the matter. 
• This deal is a joint venture with Barbara Broccoli and Michael G Wilson, both of whom have been long stewards of the bond franchise.  Amazon bought MGM studios in 2022, it gave Amazon the right to distribute bond films, while Broccoli and Wilson retained creative control. But they have been at odds with Amazon since the tech giant bought MGM studios in 2022, delaying the production of a new bond film. 
• This new deal allows them to create television and movies based on the bond character. 
• Rumors are they would like to create an expanded Bond Universe, similar to the MCU.

Closing

And that is the week in the cloud! Visit our website, the home of the Cloud Pod where you can join our newsletter, slack team, send feedback or ask questions at theCloud Pod.net or tweet at us with hashtag #theCloudPod