362: Mechanical Turk: The Fall of the Ottoman API

Welcome to episode 362 of The Cloud Pod, where the weather is always cloudy! Justin, Jonathan, and Matt are in the studio, and this week we’ve got slightly less AI, but much more data news – a trend Jonathan is sure will continue. Join us as we explore Kubernetes updates, the end of Amazon Mechanical Turk, and the last of the physical media for game consoles, plus more. There’s a lot to cover, so let’s get started! 

Titles we almost went with this week

• Going Cold Turk-ey
• Kubernetes Rollbacks Let You Ctrl-Z Your Cluster
• Answer Engine Optimization Is the New SEO Game
• Turk-ing Point: Amazon Pulls the Plug
• Stop Chasing Ephemeral IPs in Your EKS Cluster
• Firewall Rules That Know Your Pods by Name
• CloudWatch Pipelines Finally Speaks Fluent OpenTelemetry
• Sony Discs You, Keeps Your Money Forever 
• EKS Upgrades Finally Get an Undo Button
• Azure Embraces Chaos, Calls It Studio Time
• Mechanical Turk Turns Off The Lights For Good
• There Was Never Anyone Inside the Box
• KV Cache Me If You Can on GKE
• AWS Security Hub Says Hello to Azure, Finally
• Amazon Cancels the Internet’s Oldest Side Hustle
• Amazon Prime’s Next Delivery: A Pink Slip for the Turk
• Business as Usual – Azure announces Chaos Studio?
• Hello sunshine, my old Azure

A big thanks to this week’s sponsors:

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.

Follow Up 

01:36 Amazon’s Mechanical Turk to stop accepting new customers – and not  Even AI can save it

• Amazon is closing Mechanical Turk to new customers, effectively ending the crowdsourced human-labor platform that launched in 2005 and once served as a foundational tool for AI training data labeling.
• The service predates the current generative AI boom by nearly two decades, originally designed for tasks like image tagging, transcription, and data verification that computers couldn’t handle at the time.
• Ironically, the rise of AI and large language models has reduced demand for the type of human-in-the-loop micro-tasking Mechanical Turk provided, as automated systems now handle much of that labeling work.
• Existing customers can reportedly continue using the platform, but the halt on new signups signals that Amazon is deprioritizing the service rather than investing further in it.
• This closure reflects a broader shift in the AI data pipeline, where synthetic data generation and more sophisticated automated labeling tools are replacing older crowdsourced human-labor marketplaces.

02:14 Justin – “If you ever actually used the service, you’ll know that both the Turks interface that you actually did work, and the setting up the jobs was terrible anyway. It was always very difficult to use, and they never made it easier over the years. So I don’t know if they’ve ever really been investing in it heavily. But it’s definitely been around for a long time.”

AI Is Going Great – or How ML Makes Money 

04:11 New analytics and cost controls are available for Claude Enterprise

• Anthropic added richer admin analytics, model-level entitlements, and spend alerts to Claude Enterprise, giving IT and finance teams more granular visibility into how Claude is being used across groups and individual users.
• The updated analytics dashboard now breaks down usage and cost by SCIM group and user, with Claude Code getting dedicated tabs that estimate productivity lift, cost per commit, and annual value using adjustable, transparent formulas.
• An Analytics API lets finance and IT pull Claude usage and cost data into existing tools like Datadog Cloud Cost Management and CloudZero, with filtering by date range, team, product, or model, treating Claude spend like any other cloud cost line item.
• Admins can now set model defaults per role or product so routine work does not automatically route to the most expensive model, and spend-threshold alerts fire at 75% and 90% of org-level limits to prevent mid-task interruptions.
• An Admin API enables scripted cost-control workflows at scale, allowing organizations to automate spend limit reviews, flag users approaching thresholds, and monitor rapidly changing usage patterns without manual dashboard checks.

05:08 Justin – “You could put it in Datadog or in CloudZero or, you know, Cloudability, but you could also just have Claude write you a dashboard.”

06:50 Claude Cowork on web and mobile: hand off work anywhere 

• Anthropic is expanding Claude Cowork beyond desktop to mobile and web, allowing users to start tasks on one device and check or complete them on another. Beta access rolls out over several weeks, starting with Max plan subscribers.
• Cowork enables Claude to work autonomously across files, calendar, email, messaging apps, and connected tools until a task is complete, including scheduled tasks that run with no device online, such as a 6 am client briefing built from email threads and transcripts.
• Anthropic’s usage data shows over 90% of Cowork activity is non-coding work, with business operations and content creation making up roughly half of all usage, covering tasks like expense reconciliation, contract analysis, and deck building from transcripts and pipeline data.
• The approval workflow keeps humans in the loop: when Claude reaches a decision point requiring judgment, it sends a query to the user’s phone, and users can redirect work mid-task while Claude continues on the adjusted path.
• Desktop remains the full-featured version with local file and browser access, while chat and Cowork now share a unified interface with shared projects and artifacts across web and desktop; Anthropic is also doubling Cowork usage limits through August 5 for the launch.

07:15 How people are using Claude Cowork 

• Anthropic released usage data from 1.2 million sampled Claude Cowork sessions collected May 11-31, 2026, across more than 600,000 organizations, classifying activity into a 20-category taxonomy of work tasks.
• Business process and operations tasks accounted for 33.4% of usage, including reconciling spreadsheets, building onboarding checklists, and consolidating status updates, while content creation and copywriting made up 16.4%, covering drafts, slide decks, and proposals. (Please note, the copywriting is of far inferior quality to that of a human copywriter. -show note editor Heather) 
• Software development represented only 8.7% of Claude Cowork sessions, with Anthropic noting that developers tend to use Claude Code instead for core coding tasks, reserving Cowork for surrounding administrative and communication work.
• The data shows roughly half of Cowork usage falls into connective, cross-role tasks rather than core job functions, suggesting the tool is being adopted for coordination and information synthesis rather than specialized technical work.
• Anthropic gathered the data using a privacy-preserving, aggregate-only analysis method with capped hourly sampling rather than a fixed traffic percentage, meaning the reported figures represent shares of sampled sessions rather than absolute usage volumes.
• Interested in taking Anthropic’s Intro to Claude Cowork course? You can do that here. 

08:39 Jonathan – “I like the back and forth between mobile. I think one of my pain points has been starting on desktop, not necessarily with Cowork, just in general; it could be just a regular chat, as long as it’s got access to local tools. And then you sort of walk away from the computer, you go somewhere, and then it says, Hey, I finished this thing and you send it a new message, and all of a sudden it switches the tool context to the mobile device instead of the desktop, where it has access to all the files that it was working on. And so now you’re kind of in this weird limbo until you get back to your PC.”

AWS

11:08 Upgrade Amazon EKS clusters with confidence using Kubernetes version rollbacks

• AWS has added Kubernetes version rollbacks to Amazon EKS, allowing cluster administrators to reverse a minor version upgrade within a seven-day window if issues arise post-upgrade, addressing a long-standing limitation in open-source Kubernetes where control plane rollbacks were not supported.
• The feature returns clusters to their previously validated production version rather than an emulated state, and includes automated rollback readiness checks through cluster insights that flag node version compatibility and add-on dependency issues before proceeding.
• For EKS Auto Mode clusters, both the control plane and managed nodes roll back together, with the process respecting existing pod disruption budgets to maintain workload stability. A cancel API is also available if you need to stop a node rollback mid-process and adjust your approach.
• The control plane rollback process takes roughly 20 minutes, similar to a standard upgrade, and the feature is available at no additional cost across all commercial AWS regions where EKS is supported, with customers paying only standard EKS and compute costs.
• This feature is particularly relevant for organizations in regulated industries or those managing large numbers of clusters, where upgrade hesitation has led to clusters running on older versions and missing security patches due to a lack of a reliable recovery path.
• Available at no cost, you can check out version rollback on the EKS console.

14:45 Amazon CloudWatch pipelines now support processing and enriching OpenTelemetry metrics

• Amazon CloudWatch pipelines now support processing and enriching OpenTelemetry metrics during ingestion, allowing teams to add business context tags, strip high-cardinality labels, and rename metrics centrally without modifying application instrumentation.
• This addresses a common pain point where customers previously had to build custom processing layers or change source instrumentation just to transform OTel metrics before storage, which added operational overhead.
• Practical use cases include tagging metrics with team ownership or cost center data from sources you cannot modify, and reducing storage costs by removing unnecessary high-cardinality labels from custom workloads.
• The feature is available in all AWS Regions where CloudWatch pipelines and CloudWatch native OTel metrics are supported, with no additional cost for the processing itself. 
• Standard CloudWatch OTel metrics ingestion pricing still applies, so check the CloudWatch pricing page for specifics.
• To get started, navigate to the CloudWatch console under Ingestion, select Pipelines, and choose CloudWatch Metrics (OTel) as the source. Full documentation is available at the CloudWatch pipelines docs page.

15:22 Justin – “They could have just fixed the issue where you can’t add additional metadata to the tagging. That would have been a solution for this too. But okay, OTEL. Let’s do it that way.”

19:53 Amazon ECS now provides real-time deployment observability in the AWS Management Console

• Amazon ECS now offers a live deployment timeline in the console that shows each deployment phase, service events, and task launch and termination progress with automatic refresh, eliminating the need to piece together deployment status from multiple tools.
• The feature includes real-time circuit breaker status with live task failure proximity and threshold tracking, plus health checks at both the container and load-balancer level, giving teams earlier visibility into deployments heading toward failure.
• Failed tasks surface directly in the deployment timeline with diagnostic context and deep links to AWS CloudTrail, reducing the manual investigation work typically required to trace the root cause of a deployment failure.
• This is available at no additional charge for all ECS services using the rolling update deployment type, across all AWS commercial regions and AWS GovCloud (US), accessible via the Deployments tab in the ECS console.
• Teams that frequently deal with deployment failures or operate in high-velocity environments will benefit most, as the consolidated view reduces context switching and shortens the time between detecting and resolving an issue.

23:07 Enforce zero data retention on Amazon Bedrock with Bedrock Projects and service control policies

• This update tackles a real compliance headache: some new Bedrock models like Claude Fable 5 require sharing data with third-party providers, so AWS added tools to let organizations centrally block that behavior across every account, no exceptions.
• The key concept is that your configured retention mode acts as a ceiling, not a floor. Setting an account to allow provider_data_share doesn’t force sharing on every request; models supporting zero retention still operate that way regardless of the account-level setting.
• Bedrock Projects (available only on the bedrock-mantle endpoint) let teams isolate workloads with different retention needs in the same account, so a research team can experiment with data-sharing models while a production project stays locked to zero retention.
• For airtight enforcement, service control policies (SCPs) override even account administrators and root users, blocking any attempt to enable organization-wide data sharing. 
  • This works across both the bedrock-mantle and bedrock-runtime endpoints, and a single SCP at the root OU applies globally across all AWS Regions.
• One operational nuance worth flagging for listeners: cross-Region inference profiles evaluate retention mode based on the source Region of the API call, not the destination Region where processing actually happens, which matters for teams tracking data residency for compliance purposes.

24:00 Jonathan – “I feel like the past couple of years have been AI stories just dominating; I think data stories are going to be dominating the next couple of years. It’s going to be about access to data, sharing data. That’s my prediction.” 

25:00 AWS Security Hub extends unified security management to Microsoft Azure

• AWS Security Hub now extends monitoring to Microsoft Azure resources, covering VMs, Azure Container Registry images, Function Apps, and Azure identities alongside existing AWS coverage, giving multi-cloud customers a single console for security posture management rather than separate tools per cloud.
• The service checks Azure resources against CIS Benchmarks for Microsoft Azure Foundations and surfaces misconfigurations, internet exposure, and vulnerabilities using the same finding format and automation workflows as AWS findings, including existing EventBridge integrations for automated response.
• A 30-day free trial is available for Azure monitoring, starting when customers create the Azure integration; after the trial, pricing matches the cost of monitoring equivalent AWS resources.
• Azure integration can be configured from most AWS regions where Security Hub operates, though it’s currently unavailable in Middle East (UAE), Middle East (Bahrain), Asia Pacific (Taipei), and Asia Pacific (New Zealand).
• Customers can also enable Azure integrations independently through AWS Security Hub CSPM for posture checks or Amazon Inspector for vulnerability management, without requiring the full Security Hub deployment, offering flexibility for teams with narrower security tooling needs.

25:46 Justin – “Congrats, Amazon, for joining the rest of us.” 

GCP

26:45 Boost Performance and Lower Costs with AlloyDB AI Functions

• AlloyDB now includes three new GA AI functions: ai.summarize, ai.agg_summarize, and ai.analyze_sentiment, which allow developers to run sentiment analysis, text summarization, and grouped summarization directly within SQL queries without building separate data pipelines.
• Smart Batching for AI functions reduces latency and cost by deduplicating prompt overhead across rows, with Google reporting up to 2,400x performance improvement processing around 10,000 rows per second for ai.if and ai.rank, currently available in preview.
• The Optimized AI Functions feature trains a lightweight proxy model on your existing embeddings and data, allowing decisions to be processed natively inside the database rather than calling an external LLM. 
• Google reports up to 23,000x throughput improvement and cost reductions down to roughly 1/10th of a cent for qualifying workloads, though it falls back to the full LLM when accuracy thresholds are not met.
• Core AI functions ai.generate, ai.rank, ai.if, and ai.forecast are now Generally Available, making AlloyDB a more complete option for teams wanting to embed LLM capabilities directly into database queries rather than managing separate inference infrastructure.
• Practical use cases include filtering product search results by nuanced numerical constraints, consolidating customer reviews at scale, and extracting structured data from raw text, all without leaving the SQL layer. 
• Pricing is usage-based, and teams can explore the feature with a 30-day free trial.

29:15 Matthew – “I just don’t like the idea that my database is running AI queries for me.”

32:48 Scaling LLM Inference: Multi-Node KV Cache Offloading with GKE & Managed Lustre

• Google Cloud published a reference architecture for offloading LLM KV caches to Managed Lustre as a shared external filesystem tier, targeting enterprises running long-context inference workloads that exceed local CPU RAM and SSD capacity on multi-node GPU clusters.
• Benchmarks using Llama-3.3-70B on a six-node A3 Mega cluster show roughly 50% TCO reduction and nearly 60% fewer GPU-hours required, driven by a 95% cache hit rate that eliminates redundant prefill computation across nodes.
• A hybrid extension layers CPU RAM offloading on top of the Lustre tier, delivering approximately 40% improvement in Time to First Token and 30% reduction in end-to-end latency compared to CPU offload alone, which is a meaningful improvement for latency-sensitive inference serving.
• The deployment stack integrates GKE version 1.33 or later with the Managed Lustre CSI driver, vLLM, and the open-source llm-d PVC Evictor for LRU garbage collection, with validated deployment tracks for Qwen3.5-35B and Gemma 4-31B in addition to Llama.
• Operationally, the PVC Evictor requires dedicated compute resources at scale, roughly 12 CPU and 8GB memory per pod with one replica per 72 TB of Lustre capacity, so teams should factor in that management overhead when evaluating total infrastructure costs alongside Managed Lustre storage pricing.

Azure

33:57 Microsoft Frontier Company: AI engineering that amplifies and protects your intelligence

• Microsoft launched Frontier Company, a new operating business backed by a $2.5 billion investment, embedding 6,000 industry and engineering experts directly at customer sites to co-design and deploy AI systems tied to measurable business outcomes.
• The offering goes beyond traditional consulting by combining deep industry knowledge, change management, and enterprise AI engineering into a continuous improvement loop, with early deployments at organizations like LSEG, Land O’Lakes, and Novo Nordisk.
• A core principle of the model is that customer data and IP are never used to train models in ways that benefit other organizations, addressing a significant concern enterprises have raised about AI vendor relationships.
• The platform supports model flexibility across OpenAI, Anthropic, Microsoft AI, open source, and specialized industry models, meaning customers are not locked into a single provider and can select models appropriate for each workload.
• Microsoft is extending this through Global SI partners including Accenture, Capgemini, EY, KPMG, and PwC, which suggests the primary delivery channel will be partner-led rather than direct Microsoft engagement for most customers. 
• Pricing details are not publicly disclosed and would likely be scoped per engagement.

34:59 Jonathan – “The cynic in me thinks that they’re doing this as a separate company so that in eighteen months when AI can replace those forward-facing engineers, they can lay them all off without it damaging Microsoft’s own reputation as an employer. This has gotta be a this has gotta be a really short-term thing. This has gotta be information gathering.”

37:21 Proving application resilience on Azure with Chaos Studio 

• Azure Chaos Studio Workspaces is now in public preview, offering a scenario-based approach to chaos engineering that tests real-world outage patterns like Zone Down, DNS Outage, and SQL failover, rather than isolated faults. General availability is targeted for late 2026.
• The service addresses a common gap in cloud resilience: many outages stem from misconfiguration (hardcoded connection strings, misconfigured health probes) rather than platform failures, and Chaos Studio helps surface these before they hit production. 
  • This aligns with Azure’s shared responsibility model for reliability.
• Workspaces reduce setup friction by using a managed identity to auto-discover resources in a subscription or resource group and recommend applicable test scenarios, with a library of curated scenarios covering compute, database, DNS, identity, cache, and messaging failures.
• Integration with GitHub Copilot (via a dedicated Chaos Studio Skill) and a new MCP server lets engineers or AI agents provision workspaces, run drills, and pull correlated Azure Monitor signals directly from tools like Copilot, Claude, or Cursor, without needing to script against the Chaos Studio REST API directly.
• Each test run generates a structured drill report detailing injected faults, affected resources, and recovery timelines, useful for audit evidence, change tickets, or service health reviews.
• Microsoft is also positioning this as foundational for validating AI workloads (RAG pipelines, inference endpoints) and future Azure SRE agent capabilities.

39:09 Jonathan – “Microsoft is dead to me.” 

42:53 Microsoft Entra Backup and Recovery is now generally available

• Microsoft Entra Backup and Recovery has reached general availability, shifting the recovery model from point-in-time restore to a comprehensive tenant recoverability strategy. 
  • This addresses a long-standing gap in identity infrastructure protection.
• The feature is included with Entra P1 and P2 licenses, meaning organizations already on those tiers get this capability without additional cost. This lowers the barrier for adopting a more resilient identity recovery approach.
• Identity systems like Entra ID are foundational to enterprise cloud environments, so recovery capabilities for tenant-level configuration and objects reduce risk from accidental deletions, misconfigurations, or malicious changes.
• This development is relevant to disaster recovery and business continuity planning, since identity outages can cascade into broader application and access failures across an organization.
• Worth discussing how this compares to third-party backup solutions for Azure AD/Entra that have existed in the market, and whether native tooling changes the calculus for enterprises evaluating vendor lock-in versus built-in protection.

43:29 Jonathan – “How did they not have this before?” 

44:16 Microsoft’s new Azure Linux 4.0 is here, and it could replace Windows Server in the enterprise

• Microsoft released Azure Linux 4.0 as a downloadable ISO that can be installed on bare-metal servers and VMs outside of Azure, marking a shift from internal cloud plumbing to a standalone server distribution. 
  • It’s based on Fedora, uses RPMs, and ships with a hardened Linux kernel 6.18.
• Support differs significantly by deployment: running Azure Linux on Azure comes with formal SLAs, CVE patching, and integration with Defender for Cloud, while on-premises or bare-metal installs are community-supported only, with no official Microsoft backing.
• Linux has been the dominant OS on Azure for nearly a decade, so this move formalizes and extends that reality by giving Microsoft its own distribution to compete with AlmaLinux, Rocky Linux, and other enterprise Linux options.
• The article raises the question of whether Windows Server’s long-term relevance is diminishing as Microsoft invests more in its own Linux distribution, a point worth discussing given Microsoft’s historical Windows Server ecosystem.
• Worth discussing: at this early beta stage, Azure Linux lacks the maturity and features of established Red Hat-based distros, so enterprise adoption outside Azure may be limited until the platform matures.

44:54 Justin – “At the end of the day, is this thing gonna replace Windows Server? Probably not. ZDNet seems to think it will, but I don’t think so. And the only thing I’m gonna use this for is probably to go into my WLM on my Windows box when I rarely need Window an Ubuntu prompt on my Windows box.”

Emerging Clouds

48:19  Your site, your rules: new AI traffic options for all customers

• Cloudflare is expanding its AI bot controls beyond a simple block-all toggle, introducing three distinct categories for all customers, including the free tier: Search bots that index content for later retrieval, Agent bots acting on behalf of users in real time, and Training bots that absorb content into model weights.
• Starting September 15, 2026, new domains on Cloudflare will have Training and Agent bots blocked by default on ad-supported pages, while Search remains allowed, with multi-purpose crawlers like Googlebot and BingBot subject to the most restrictive applicable rule if they also perform training.
• BotBase is a new Enterprise Bot Management feature providing a searchable database of all verified bots with classification details, detection IDs for use in security rules, and planned expansion into a direct control center for managing automated traffic.
• Cloudflare is introducing a content use signal extending robots.txt via Content Signals, with three levels: immediate meaning store nothing, reference meaning index and link back, and full meaning summarize and reproduce, allowing site owners to express granular preferences beyond simple allow or block.
• The transitive trust proposal uses the existing RFC 7239 Forwarded header to carry bot operator identity and content use declarations through multiple proxy layers, with the incentive that losing verified status across the roughly 20 percent of web domains behind Cloudflare serves as a meaningful deterrent against abuse.

48:48 Content Independence Day, one year on: building the business model for the agentic Internet

• One year after Cloudflare blocked AI training crawlers by default for new domains, over 50% of internet traffic is now non-human, and AI training crawlers account for 52% of all crawler requests as of June 2026, up from 22% in Spring 2025.
• The traditional search-to-referral traffic model is breaking down, with some heavily crawled industries seeing human traffic decline as much as 40% in under a year, pushing publishers toward what they call “Google Zero” planning.
• Cloudflare calls out Google specifically for running a mixed-use crawler that combines search indexing and AI training into a single bot, giving Google access to roughly 2x more content than competing AI companies while preventing publishers from separating consent for each purpose.
• More than 50 publisher-AI licensing agreements have been signed since 2023, but Cloudflare notes these deals remain bespoke and are unlikely to fully replace lost referral and advertising revenue, leaving content valuation largely unresolved.
• Cloudflare is positioning itself as infrastructure for this emerging content economy, noting that over one-third of crawler activity on its network still comes from mixed-use bots, and stating a goal of driving that number to zero within the next year through new attribution and signaling tools.

49:07 Making AI search smarter

• Cloudflare is launching a research program to help AI search engines identify fresh, high-quality content without redundant crawling. 
• Their data shows over 50% of crawl traffic from good bots re-fetches pages that have not changed, creating unnecessary costs for site owners.
• A 2025 Pew Research study cited in the article found that when Google shows an AI summary, users click traditional search links only 8% of the time, down from roughly 16% without a summary. This illustrates the direct revenue threat AI search poses to content publishers.
• Cloudflare is evolving its Pay Per Crawl model toward Pay Per Use, running experiments with partners like Ceramic.ai and You.com. 
  • Ceramic uses a pay-per-query approach where publishers get compensated each time their content appears in search results, not just when it is crawled.
• The program includes new reporting tools for content owners, covering top queries driving their content into AI results, specific snippets used, and average ranking positions. Cloudflare is framing this as answer engine optimization, a parallel to traditional SEO.
• Cloudflare is positioning itself as a neutral infrastructure layer for this emerging content compensation market, covering more than 20% of the web. The company states that no content is shared for foundation model training, and participation is limited to search use cases.

49:25 Announcing the Monetization Gateway: charge for any resource behind Cloudflare via x402

• Cloudflare is launching a Monetization Gateway that lets customers charge for any asset behind their network, including APIs, datasets, web pages, and MCP tools, using usage-based pricing enforced at the edge rather than at the origin.
• The system is built on x402, an open protocol that uses the long-dormant HTTP 402 status code to handle payment negotiation inline within standard HTTP requests, with no redirects, no account creation, and settlement in stablecoins like USDC in under a second.
• A key technical benefit is that payment verification and enforcement happen on Cloudflare’s global network across 330+ cities, which means origin servers never handle payment logic or absorb the traffic load from high-volume payment verification.
• The model is specifically designed for AI agents as buyers, since agents can make thousands of micropayments autonomously without human approval friction, and the x402 protocol supports sub-cent transactions that would be economically unviable on traditional payment rails.
• Businesses can configure payment rules through a dashboard, API, or Terraform, allowing them to charge per route, per HTTP verb, or only for unauthenticated callers, with the option to layer in Web Bot Auth for identity verification alongside payment. 

After Show 

51:28 Sony announces end of PlayStation discs, parts of digital store in the sameday – Ars Technica

• Sony announced it will stop producing physical PlayStation game discs in January 2028, citing that digital downloads now account for 78 percent of full-game unit purchases in its most recent fiscal year.
• The shift completes a move to a licensing-only model, which is a distinction worth noting: customers purchasing digital games are buying a personal, non-transferable license rather than owning the product outright, per Sony’s own terms of service.
• This mirrors patterns cloud professionals see regularly with SaaS and digital services, where vendors retain control over access and can revoke or alter it, raising questions about long-term consumer rights and data portability.
• A historical precedent from 2013 exists where Valve removed purchased games from customer libraries after a server shutdown, illustrating that the risk of losing access to paid digital content is not purely theoretical.
• For the broader tech industry, this move reinforces the ongoing tension between convenience-driven digital distribution and consumer ownership, a conversation that extends well beyond gaming into software, media, and cloud-hosted services generally.

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