About this episode
Welcome to episode 313 of The Cloud Pod, where your hosts, Matt, Ryan, and Justin, are here to bring you all the latest in Cloud and AI news. This week we’ve got an installation of Cloud Journey featuring Gartner and chaos AND an aftershow! We’ve got acquisition news, new tools, an undersea cable, and even a little chaos, all right now in the cloud. Let’s get into it! Titles we almost went with this week: From Vibe Check to Production Spec Node More Mr. Nice Guy: AWS Locks Down Access Until You Ask Nicely Grok’s New Feature: Ask Elon First The AI That Phones Home to Dad Musk-See TV: When Your Chatbot Needs Parental Guidance Oracle’s Federal Discount: 75% Off for Six Months (Terms and Conditions Apply) GameDay: Not Just for Sports Anymore Bob the Builder Center: Can We Fix AWS? Yes We Can! Bucket List: Google Cloud Storage Finally Lets You Pack Up and Move The Great Bucket Migration: No Forwarding Address Required Compose Yourself: Cloud Run Gets Docker-mented Survey Says: Your Team Needs a Performance Check-Up From Florida With Love: Google’s New Cable Has a License to Transmit Sol Train: Google Lays Track Across the Atlantic Finding the Right Gradient for Your AI Journey Google Cracks the Code on AWS’s Cloud Castle Breaking Cloud: Google’s Data Analytics Cook Up Market Share From Chat to Churn: The Great GPT Subscription Exodus AWS Finally Filters Out the Pricing Noise The Price is Right: AWS Edition Gets New Search Features Four Filters and a Pricing API Walk Into a Cloud Fee-fi-fo-fum who has a flash reasoning model Follow Up 02:01 Cognition to buy AI startup Windsurf days after Google poached CEO Cognition acquired Windsurf’s IP, product, and remaining talent after Google hired away the CEO and senior staff, highlighting the intense competition for AI coding expertise among major tech companies. The deal follows a failed $3 billion acquisition attempt by OpenAI and Google’s $2.4 billion licensing and compensation package to secure Windsurf’s leadership, demonstrating the premium valuations for AI coding technology. Both companies develop AI coding agents designed to accelerate software development, with Cognition’s Devin agent and Windsurf’s tools representing the growing market for AI-powered developer productivity solutions. The acquisition ensures all Windsurf employees receive accelerated vesting and financial participation, addressing the disruption caused by the leadership exodus to Google. This consolidation in the AI coding space suggests smaller startups may struggle to retain talent and remain independent as tech giants aggressively pursue AI engineering capabilities. AI Is Going Great – Or How ML Makes Money 04:40 New Grok AI model surprises experts by checking Elon Musk’s views before answering – Ars Technica Grok 4 , xAI’s latest AI model, has been observed searching for Elon Musk’s X posts when answering controversial questions, with the model’s reasoning trace showing searches like “from:elonmusk (Israel OR Palestine OR Gaza OR Hamas)” before formulating responses. The behavior appears inconsistent across users and prompts – while some see Grok searching for Musk’s views, others report the model searching for its own previous stances or providing different answers entirely. This discovery highlights potential challenges in AI alignment and bias in cloud-hosted LLMs, where models may inadvertently incorporate owner preferences into their decision-making processes without explicit programming. The SuperGrok tier costs $22.50/month and includes visible reasoning traces similar to OpenAI’s o3 model , allowing users to see the model’s search queries and thought process during response generation. For cloud providers and enterprises deploying AI services, this raises important questions about model transparency, bias detection, and the need for robust testing frameworks to identify unexpected behaviors before production deployment. 06:23 Ryan – “It’s all my concerns about the bro-coders and the culture and Musk’s cult of personality dictating things, and not being something that can be trusted.” 06:53 Introducing GradientAI: DigitalOcean’s Unified AI Cloud | DigitalOcean DigitalOcean launches GradientAI , a unified AI cloud platform that combines GPU infrastructure, agent development tools, and pre-built AI applications into a single integrated experience for the full AI development lifecycle. The platform consists of three main components: Infrastructure (GPU compute for training/inference), Platform (agent development environment), and Applications (pre-built AI agents for common use cases like customer support). New GPU options are being added, including AMD Instinct MI325X (available this week) and NVIDIA H200s (next month), providing more choice and performance options for both training and inference workloads. The Platform component will support Model Context Protocol (MCP) , multi-modal capabilities, agent memory, and framework integrations to simplify moving AI projects from prototype to production. This positions DigitalOcean to compete more directly with major cloud providers in the AI space by offering a simpler, more integrated alternative for digital native enterprises building AI applications. 07:42 Ryan – “I’m in support of any feature that Digital Ocean puts on their cloud, just because I’m rooting for the underdog there. And if you are a Digital Ocean customer, how great is it to have this and not to go to one of the other cloud hyperscalers and maintain two separate infrastructures?” 09:07 Companies Canceling ChatGPT Subscriptions Companies are canceling ChatGPT subscriptions due to concerns about data security, cost-benefit analysis, and integration challenges with existing enterprise systems. Organizations report difficulty justifying the $20-30 per user monthly cost when employees use the tool sporadically or for non-critical tasks. The trend highlights a growing enterprise preference for self-hosted or private cloud AI solutions that offer better data governance and compliance controls. Companies are exploring alternatives like Azure OpenAI Service or AWS Bedrock that integrate with existing cloud infrastructure and security policies. Technical teams cite API limitations, lack of fine-tuning capabilities for domain-specific tasks, and inability to train on proprietary data as key factors driving cancellations. Many organizations need models that can be customized for industry-specific terminology and workflows. The shift suggests enterprises are moving from experimental AI adoption to more strategic implementation focused on measurable ROI and specific use cases. Companies are consolidating around platforms that offer both general-purpose and specialized models within their existing cloud environments. This development indicates a maturing AI market where businesses demand enterprise-grade features like audit trails, role-based access control, and integration with existing identity management systems rather than standalone consumer-oriented tools. 10:23 Justin – “I know I cancelled my ChatGPT subscription months ago; I was a trend setter.” Cloud Tools 13:53 2025 DORA Survey is now open | Google Cloud Blog The 2025 DORA survey is now open until July 18, offering teams a 10-15 minute self-assessment tool to benchmark their software delivery and operational performance against industry standards. This year’s survey focuses heavily on AI adoption across the software development lifecycle, with 76% of technologists already using AI in their daily work. Companies applying DORA principles have achieved dramatic improvements – Banorte increased deployment frequency from bi-weekly to multiple times daily , SLB cut deployment time from 5 days to 3 hours , and GitLab reduced errors by 88% . These metrics demonstrate the tangible value of continuous improvement practices backed by data-driven insights. The survey explores how organizations can maximize AI impact while maintaining developer well-being, finding that transparent AI strategies and governance policies significantly increase adoption rates. It also examines trust in AI systems and how teams can best support the transition to AI-enhanced workflows. Available in 6 languages, the survey welcomes input from all software delivery roles – engineers, product managers, CISOs, and UX designers – to capture diverse perspectives on team performance. Participants gain immediate value through structured reflection on their workflows and bottlenecks. DORA’s research continues to shape industry understanding of high-performing teams, with findings like the substantial impact of quality documentation on team performance. The anonymous data collected helps establish benchmarks and best practices for the entire technology community. Listener note: The survey is now closed, so all arguments about the closing date are moot. We will bring you the results of said survey as soon as they’re released. AWS 17:02 Introducing Just-in-time node access using AWS Systems Manager | AWS Cloud Operations Blog Yes, we originally missed this one. But maybe you’ve seen it in the console, just like Matt. AWS Systems Manager now offers just-in-time node access, enabling temporary, policy-based access to EC2, on-premises, and multicloud nodes without maintaining long-term credentials or SSH keys. This addresses the security vs operational efficiency trade-off many organizations face when managing thousands of nodes. The feature supports both manual approval workflows (with multiple approvers) and automated approval using Cedar policy language, allowing organizations to implement zero standing privileges while maintaining rapid incident response capabilities. Access automatically expires after a defined time window. Integration with Slack , Microsoft Teams , and email notifications streamlines the approval process, while EventBridge events enable audit trails and custom workflows. Sessions can be logged for commands and RDP recordings for compliance requirements. AWS offers a free trial period covering the remainder of the current billing period plus the entire next billing period per account per Region, after which pricing is usage-based. This allows organizations to test configurations and policies before committing to costs. The solution works seamlessly across AWS Organizations, supporting consistent access controls whether managing single or multiple accounts, with administrators defining policies, operators requesting access, and approvers managing requests through a unified console experience. 18:36 Matt – “It runs on Jonathan’s favorite method of security, which is through tags. So a lot of the automation, a dev person can automatically get access if tag equals dev is in there. So, there are some features or setup design of it that might not be what works for your company, but there is some like prep work if you want to use it, but it does seem like a really nice feature.” 25:11 Introducing Kiro – Kiro Kiro is a new AI-powered IDE that introduces spec-driven development, automatically generating requirements, technical designs, and implementation tasks from simple prompts to help developers move from prototype to production-ready applications. The platform’s key innovation is its specs feature, which creates EARS notation acceptance criteria , data flow diagrams, TypeScript interfaces, and database schemas that stay synchronized with the evolving codebase, addressing the common problem of outdated documentation. Kiro hooks provide automated quality checks by triggering AI agents on file events – for example, automatically updating test files when React components change or scanning for security vulnerabilities before commits, enforcing consistent standards across development teams. Built on Code OSS with VS Code compatibility, Kiro supports Model Context Protocol for specialized tool integration and is currently free during preview with some limitations, targeting developers who need more structure than typical AI coding assistants provide. This represents a shift toward more structured AI-assisted development, moving beyond simple code generation to address production concerns like maintainability, documentation, and team consistency that traditional AI coding tools often overlook. 26:19 Justin – “I’ve been playing with it most of the day, building a mobile app across platform, which I’ve never done before, and I have no experience doing and I have no idea what it’s doing. But, it’s working great.” 35:00 New Amazon EC2 P6e-GB200 UltraServers accelerated by NVIDIA Grace Blackwell GPUs for the highest AI performance | AWS News Blog AWS launches P6e-GB200 UltraServers with NVIDIA Grace Blackwell GPUs , offering up to 72 GPUs in a single NVLink domain with 360 petaflops of FP8 compute and 13.4 TB of HBM3e memory for training trillion-parameter AI models. The new instances use NVIDIA’s superchip architecture that combines Blackwell GPUs with Grace ARM CPUs on the same module, providing significantly higher GPU-CPU bandwidth compared to current P5en instances while delivering 28.8 Tbps of EFA networking. P6e-GB200 UltraServers are only available through EC2 Capacity Blocks for ML in the Dallas Local Zone (us-east-1-dfw-2a), requiring upfront payment for reserved capacity blocks of either 36 or 72 GPUs with pricing determined at purchase time. Integration with AWS services includes SageMaker HyperPod for managed infrastructure with automatic fault replacement within the same NVLink domain, EKS with topology-aware routing for distributed workloads, and FSx for Lustre , providing hundreds of GB/s throughput for large-scale AI training. The instances target frontier AI workloads, including a mixture of expert models, reasoning models, and generative AI applications like video generation and code generation, positioning AWS to compete in the high-end AI infrastructure market. 36:14 Ryan – “So if you’re a big enough Amazon customer, you can get Amazon to run your Amazon outpost with custom hardware. Cool!” 37:29 Introducing AWS Builder Center: A new home for the AWS builder community | AWS News Blog AWS Builder Center consolidates developer resources from AWS Developer Center and community.aws into a single platform at builder.aws.com , providing a unified hub for accessing tutorials, workshops, and community engagement tools. The new Wishlist feature allows developers to submit and vote on feature requests for AWS services, giving the community direct input into product roadmaps and enabling AWS teams to prioritize development based on actual user needs. Built-in localization supports 16 languages with on-demand machine translation for user-generated content, removing language barriers for global collaboration among AWS builders and expanding accessibility to non-English speaking developers. The platform integrates AWS Builder ID for consistent profile management across all AWS services, offering personalized profiles with custom URLs and QR codes for networking at events and conferences. Connect features highlight AWS Heroes , Community Builders , User Groups , and Cloud Clubs , making it easier to find local meetups and connect with experts in specific AWS service areas or technologies. 39:32 AWS Price List API now supports four new Query Filters – AWS AWS Price List Query API adds four new filter types, enabling exact attribute matching, substring searches, and include/exclude lists for more targeted product searches across AWS services. The update simplifies finding specific product groups like all m5 EC2 instance types with a single filter instead of multiple complex queries, reducing API calls and improving efficiency. This enhancement addresses a common pain point for cost optimization tools and FinOps teams who need to programmatically analyze AWS pricing data across thousands of SKUs. The new filters are available in all regions where the Price List API is supported, making it easier for organizations to build automated pricing analysis and comparison tools. Real-world applications include building custom cost calculators, automated pricing alerts, and multi-region price comparison tools for Reserved Instance planning. 40:25 Justin – “AWS CLI filtering is one of those things that drives me crazy, because I never really remember it properly. And it brings me such joy to watch the AI Bots screw it up. If the AI bot who has the documentation in its brain memorized can’t get this right, I don’t feel so bad.” 42:17 Announcing Model Context Protocol (MCP) Server for AWS Price List – AWS AWS releases an open-source Model Context Protocol (MCP) server that gives AI assistants like Amazon Q Developer CLI and Claude Desktop direct access to AWS pricing data, including on-demand, reserved, and savings plan options across all regions. The MCP server enables natural language queries about AWS pricing and product availability, allowing developers to ask questions like “What’s the cheapest EC2 instance for machine learning in us-east-1?” and get real-time responses from the AWS Price List API . This addresses a common pain point where engineers manually navigate complex pricing pages or write custom scripts to compare costs across services and regions, and now AI assistants can handle these queries instantly. The server uses standard AWS credentials and minimal configuration, making it straightforward to integrate into existing workflows where teams already use AI assistants for development tasks. Available now in the AWS Labs GitHub repository at no additional cost beyond standard AWS Price List API usage. 43:09 Matt – “When was the last time you had an engineer (or developer) go in to figure out what EC2 instance type they should use? Because everyone I’ve met just goes ‘ooh, this one’s big and shiny, we’ll put more power behind it, and that makes my code go faster’….don’t worry about your CFO’s brain exploding on the other side of it. ” 45:23 Amazon DocumentDB (with MongoDB compatibility) introduces support for up to 10 secondary Region clusters – AWS Amazon DocumentDB Global Clusters now supports up to 10 secondary regions, doubling the previous limit of 5, enabling broader geographic distribution for applications requiring low-latency reads across multiple continents. This expansion addresses disaster recovery needs by allowing organizations to replicate their MongoDB -compatible workloads across more AWS regions, reducing the blast radius of regional outages while maintaining local read performance. The increased region support particularly benefits global enterprises running customer-facing applications that need to comply with data residency requirements across multiple jurisdictions while maintaining consistent performance. While the feature enhances availability and global reach, customers should consider the cost implications of running clusters across 10 regions, including cross-region data transfer charges and compute costs for each regional cluster. This positions DocumentDB more competitively against MongoDB Atlas , which supports similar multi-region deployments, giving AWS customers a fully managed alternative without leaving the AWS ecosystem. 47:24 Amazon SageMaker Studio now supports remote connections from Visual Studio Code – AWS SageMaker Studio now allows developers to connect their local VS Code installations directly to SageMaker’s managed compute resources, reducing setup time from hours to minutes while maintaining existing security boundaries. Developers can authenticate through the AWS Toolkit extension or SageMaker Studio’s web interface, then access their SageMaker development environments with a few clicks while keeping their preferred VS Code extensions and AI-assisted development tools. This addresses a common friction point where data scientists want their familiar local IDE setup but need access to scalable cloud compute and datasets stored in AWS without complex SSH tunneling or VPN configurations. The feature complements SageMaker Studio’s existing JupyterLab and Code Editor options, giving teams flexibility to choose between web-based or local development experiences while leveraging the same underlying infrastructure. Currently available only in US East (Ohio) region, suggesting this is an early rollout that will likely expand to other regions based on customer adoption and feedback. 48:25 Ryan – “It’s definitely kept me from adopting SageMaker, and a larger thing being sort of forced into their interface and their notebook interface. I do like it locally. It wasn’t terrible; I could use it before, but it’s a lot easier if I don’t have to do that. So I like that this pattern is becoming more prevalent, where you’re keeping your context focused directly in that IDE and the IDEs are going and reaching out to the different services.” GCP 50:16 Backup for GKE supports cross-project backup and restore | Google Cloud Blog Backup for GKE now supports cross-project backup and restore in preview, allowing users to back up workloads from one Google Cloud project, store them in a second project, and restore to a third project. This addresses a key challenge in multi-project GKE deployments where teams need centralized backup management across project boundaries. The feature enables critical disaster recovery capabilities by storing backups in separate projects and regions, protecting against regional outages or compromised primary projects. Organizations can meet RTO/RPO objectives while simplifying regulatory compliance through proper backup isolation. Cross-project functionality streamlines development workflows by enabling easy environment seeding and cloning – teams can populate staging environments with production backup data or create isolated sandboxes without complex manual processes. Developers can be granted Delegated Restore Admin roles to restore specific backups without accessing live production environments. This positions GCP competitively with AWS and Azure backup solutions that already support cross-account/subscription backup scenarios. The integration with GKE’s existing backup infrastructure means no additional tools are required beyond configuring backup and restore plans to point to different projects. Access to the preview requires completing a form, which can be found here . No specific pricing changes were mentioned, suggesting it uses existing Backup for GKE pricing models. 51:54 Introducing Cloud Storage bucket relocation | Google Cloud Blog Google Cloud Storage introduces bucket relocation , the first feature among major cloud providers that allows moving storage buckets to different regions without changing bucket names or disrupting applications. This preserves all metadata, including storage classes, timestamps, and permissions, while maintaining object lifecycle management rules. The feature uses asynchronous data copying to minimize downtime during migration, with only a brief write-lock period during final synchronization. Organizations can perform dry runs to identify potential issues like CMEK incompatibilities before initiating the actual move. Key use cases include improving data locality for performance, meeting regional compliance requirements, and optimizing costs by moving between storage tiers. Spotify and Groupon have reported successful migrations of petabytes of data with minimal manual effort compared to traditional approaches. Bucket relocation is part of Google’s Storage Intelligence suite and supports moves between regional, dual-region, and multi-region configurations. The three-step process (dry run, initiate relocation, finalize) can be completed through simple gcloud commands. This addresses a significant pain point in cloud storage management, where previously, organizations had to use Storage Transfer Service to copy data to new buckets with different names, requiring application updates and risking extended downtime. 34:06 Matt – “This is a really cool feature that would have saved me much time in the past life of, hey, we set up this thing years before we actually started using the cloud, and it was for this one thing, and now we’ve launched everything in this other region. And every time we have to access this one specific bucket, it is somewhere else. And how do we fix that? And their process is pretty cool, too, where it sets it up, does the sync, and sits at 99% and you do that last one. This is a great quality of life feature.” 55:20 Cloud Run and Docker collaboration | Google Cloud Blog Cloud Run now supports direct deployment of Docker Compose files through the new gcloud run compose up command, eliminating manual infrastructure translation between local development and cloud deployment. This private preview feature automatically builds containers from source and leverages Cloud Run’s volume mounts for data persistence. The integration supports Docker’s new models attribute in the Compose Specification , enabling developers to deploy AI applications with self-hosted LLMs and MCP servers using a single configuration file. This positions Cloud Run as a cost-effective option for AI workloads with pay-per-second billing and scale-to-zero capabilities. Cloud Run GPUs (now generally available) combined with Compose support creates a streamlined path for AI development, with approximately 19-second time-to-first-token for models like gemma3:4b. This competes directly with AWS App Runner and Azure Container Apps but with native GPU support. The collaboration addresses the growing complexity of agentic AI applications by supporting Docker’s MCP Gateway and Model Runner, allowing developers to maintain consistent configurations across local and cloud environments. Sign up for private preview at https://forms.gle/XDHCkbGPWWcjx9mk9. This positions GCP strategically in the AI infrastructure market by adopting open standards (Compose Specification) while leveraging Cloud Run’s existing strengths in serverless compute, making it practical for teams already using Docker Compose who need GPU-accelerated AI deployments without infrastructure management overhead. Want to sign up for the private preview? You can do that here . 56:62 Ryan – “I’m curious to see the rough edges on this because you’ve been able to do sort of continuous integration delivery with CloudRun for a while, but it had to be a publicly available Github Repo, so I’m hoping that this is as transparent as it’s made to be.” 57:26 Announcing Sol transatlantic cable | Google Cloud Blog Google announces Sol, a new transatlantic subsea cable connecting the U.S. (Palm Coast, Florida), Bermuda, the Azores, and Spain (Santander), marking the first operational fiber-optic cable between Florida and Europe. This complements their existing Nuvem cable to create redundant transatlantic paths with terrestrial interconnections at multiple points. The cable strengthens Google Cloud’s global infrastructure across 42 regions by providing increased capacity, improved reliability, and reduced latency for AI and cloud services between the Americas and Europe. Sol features 16 fiber optic cable pairs and will be manufactured in the U.S. Google is partnering with DC BLOX for the Florida landing station and developing a terrestrial route to their South Carolina cloud region , while Telxius provides infrastructure in Spain to integrate with the Madrid cloud region . This positions Florida and Spain as new connectivity hubs for Google’s network. Sol joins Google’s growing subsea cable portfolio, including Nuvem, Firmina, Equiano, and Grace Hopper, demonstrating their continued investment in owning network infrastructure rather than relying solely on consortium cables. This gives Google more control over capacity, routing, and performance for its cloud customers. The cable addresses growing demand for transatlantic connectivity driven by AI workloads and cloud adoption, while also providing economic benefits to landing locations through job creation and positioning them as digital hubs. No specific cost or availability timeline was provided in the announcement. Also, we all agree this is a terrible diagram. Genuinely – the worst one we’ve seen in a while. 1:00:33 Google Finds a Crack in Amazon’s Cloud Dominance Google is gaining ground in cloud market share by focusing on data analytics and AI workloads, areas where they have technical advantages over AWS through services like BigQuery and Vertex AI. The company has shifted strategy from trying to match AWS feature-for-feature to emphasizing their strengths in machine learning infrastructure and data processing capabilities that leverage their search and AI expertise. Google Cloud’s growth rate now exceeds both AWS and Azure, though from a smaller base, with particular success in industries like retail and financial services that need advanced analytics. Key differentiators include BigQuery’s serverless architecture that eliminates capacity planning and Vertex AI’s integration with Google’s pre-trained models, making enterprise AI adoption more accessible. The strategy appears to be working with notable customer wins, including major retailers and banks, who cite Google’s superior data analytics performance and lower total cost of ownership for specific workloads. 1:01:31 Ryan – “It is interesting because I will say that this is focusing on Google’s strengths, and I agree that containers have been a strength for a long time. And you start adding BigQuery and Vertex AI, you’ve got a pretty powerful platform to build off of. The feature-to-feature, it’s going to miss all those enablements that make it really easy to stand up a full application on the cloud. So, like it’s kind of a bummer, but we’ll see what it’s actually like.” Azure 1:02:52 Reasoning reimagined: Introducing Phi-4-mini-flash-reasoning | Microsoft Azure Blog Microsoft introduces Phi-4-mini-flash-reasoning, a 3.8B parameter model using a new decoder-hybrid-decoder architecture called SambaY that combines Mamba state space models with sliding window attention and gated memory units to achieve 10x higher throughput and 2-3x latency reduction compared to standard transformer models. The model targets edge computing and resource-constrained environments where compute, memory, and latency are critical factors, making it deployable on a single GPU while maintaining advanced math reasoning capabilities with 64K token context length. Key innovation is the Gated Memory Unit (GMU) mechanism that enables efficient layer representation sharing, preserving linear prefilling time complexity while improving long-context retrieval performance for real-time applications. Primary use cases include on-device reasoning assistants, adaptive learning platforms, and interactive tutoring systems that require fast logic inference, with the model available on Azure AI Foundry , NVIDIA API Catalog , and Hugging Face . The architecture represents a practical approach to deploying AI reasoning capabilities at the edge without cloud dependency, addressing the growing need for low-latency AI inference in mobile and IoT applications. 1:04:361 Matt – “I think it’ll be interesting when you’re on your mobile device and you say, hey, run me this thing, it tries to run it on a model like this, and then if it can’t get you a good result because it’s not enough data points and parameters, then it kind of goes off. So that’s kind of where I see this going, which is edge-based computing kind of coming back alive, where your phone and your laptop, everything else has enough that could run these small models to give you, you know, just quick feedback and do it offline also, versus everything always having to happen to be online.” Oracle 1:06:43 Oracle Cloud Cuts Costs and Propels Missions for Government Agencies Oracle partnered with GSA to offer federal agencies 75% discounts for six months on licensed technologies plus migration services to Oracle Cloud, targeting the significant number of government systems still running older Oracle versions on-premises. Oracle claims its second-generation cloud offers 50% lower compute costs, 70% lower storage costs, and 80% lower networking costs compared to competitors, though these comparisons lack specific benchmarks or competitor names. The partnership removes data egress fees when moving workloads between FedRAMP and DOD IL4/IL5 certified clouds, addressing a common pain point for government agencies considering multi-cloud strategies. Oracle is positioning its integrated AI capabilities in Database 23ai and application suites as differentiators, though the announcement provides no technical details about actual AI features or how they compare to AWS, Azure, or GCP offerings. While Oracle emphasizes cost savings and modernization benefits, the real impact depends on how many federal agencies migrate from their legacy Oracle systems, which have persisted precisely because Oracle doesn’t force upgrades. Here’s the gotcha: the discounts don’t last forever. Cloud Journey 1:08:31 4 Chaos Engineering recommendations from Gartner Gartner’s 2025 Hype Cycle for Infrastructure Platforms highlights Chaos Engineering as essential for testing AI resilience, particularly for applications using generative AI API calls that need validated fallback patterns when services fail or experience latency GameDays are becoming critical for enterprise preparedness against catastrophic failures like CrowdStrike or cloud provider outages, with financial institutions using them to verify disaster recovery plans for operational resilience compliance Organizations should prioritize Chaos Engineering on business-critical systems first, focusing on payment services, single points of failure, and elevated security privilege components, where downtime costs average $14,056 per minute Reliability scoring platforms provide measurable metrics beyond simple uptime/downtime tracking, enabling teams to identify performance degradation and latency issues before they impact users The increasing complexity of modern systems combined with AI adoption makes proactive reliability testing through Chaos Engineering a necessity rather than optional, as outages cost Global 2000 companies an average of $200 million annually. After Show 1:13:02 Stop forcing AI tools on your engineers – by Anton Zaides Engineering managers face pressure to force AI tool adoption on teams, but mandating specific tools like Cursor or requiring token usage metrics can backfire and slow productivity rather than improve it Companies should give engineers dedicated time (20% workload reduction or full exploration weeks) to experiment with AI tools in their actual codebases rather than expecting zero-cost adoption The focus should shift from measuring AI tool usage to measuring actual outcomes – if engineers using AI tools deliver better results, share those specific workflows internally rather than generic success stories Monday.com’s approach of a 5-week AI exploration with 127 internal demo submissions shows how large organizations can enable organic adoption through peer-led workshops and real use case sharing AI tools excel in greenfield projects and simple codebases, but adapting them to complex existing systems requires careful evaluation of what actually works versus following industry hype. 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 theCloudPod.net or tweet at us with the hashtag #theCloudPod