Cloud Computing Solutions

We're introducing hybrid compute for all users of the Perplexity Mac app. This will allow Computer to orchestrate local models that can run locally on Mac, particularly for agent steps involving sensitive and private files (eg your bloodwork, tax returns, litigation, etc). Most AI agent apps don't differentiate between native and web-based runtimes when using cloud-based agents. Mac offers a big opportunity to move token consumption to Apple Silicon (with no price paid for tokens consumed locally) and protect user privacy. Hybrid compute combines the best of cloud-based frontier models and a privacy-protecting local runtime. We're also open-sourcing the PII classifier that we use for deciding when to send the workload to the local model in the hybrid compute setup. HuggingFace: https://lnkd.in/gF4jVvDJ Research Blog: https://lnkd.in/gDpd_GSs Product Page: https://lnkd.in/gZEFg4rM

One of the most important laws of frugal architecture is that you can’t optimize what you can’t measure. I learned this long before cloud computing. Growing up in Amsterdam during the energy crisis of the 1970s, we had things like car-free Sundays and rationed energy, but the detail that always stuck with me was closer to home. Households with their energy meter on the main floor of their homes used significantly less energy than those with it hidden in the basement. The same style of house, in the same city, yet dramatically different behaviour. About as clear of a signal as you can get that seeing data changes what you do with it. For years, in the absence of better sustainability metrics, usage (or consumption) was the best proxy we had. The meter was in the basement. With the AWS Sustainability Console, we bring the meter to your “living room”. It gives your builders direct access to Scope 1, 2, and 3 emissions data, broken down by service and Region, exportable via API, without ever touching sensitive cost and billing data. The right data, to the right people, through the right door. When carbon emission becomes just another metric in your observability stack sitting next to latency, cost, and error rates, it stops being a compliance exercise and starts becoming an architectural discipline. The world we are building in the cloud is the world we are leaving to our children. Measure it like it matters. Read more here: https://lnkd.in/efFjU7hG

Every cloud provider faces the same AI infrastructure challenge: chips need to be positioned close together to exchange data quickly, but they generate intense heat, creating unprecedented cooling demands. We needed a strategic solution that allowed us to use our existing air-cooled data centers to do liquid cooling without waiting for new construction. And it needed to be rapidly deployed so we could bring customers these powerful AI capabilities while we transition towards facility-level liquid cooling. Think of a home where only one sunny room needs AC, while the rest stays naturally cool – that’s what we wanted to achieve, allowing us to efficiently land both liquid and air-cooled racks in the same facilities with complete flexibility. The available options weren't great. Either we could wait to build specialized liquid-cooled facilities or adopt off-the-shelf solutions that didn't scale or meet our unique needs. Neither worked for our customers, so we did what we often do at Amazon… we invented our own solution. Our teams designed and delivered our In-Row Heat Exchanger (IRHX), which uses a direct-to-chip approach with a "cold plate" on the chips. The liquid runs through this sealed plate in a closed loop, continuously removing heat without increasing water use. This enables us to support traditional workloads and demanding AI applications in the same facilities. By 2026, our liquid-cooled capacity will grow to over 20% of our ML capacity, which is at multi-gigawatt scale today. While liquid cooling technology itself isn't unique, our approach was. Creating something this effective that could be deployed across our 120 Availability Zones in 38 Regions was significant. Because this solution didn't exist in the market, we developed a system that enables greater liquid cooling capacity with a smaller physical footprint, while maintaining flexibility and efficiency. Our IRHX can support a wide range of racks requiring liquid cooling, uses 9% less water than fully-air cooled sites, and offers a 20% improvement in power efficiency compared to off-the-shelf solutions. And because we invented it in-house, we can deploy it within months in any of our data centers, creating a flexible foundation to serve our customers for decades to come. Reimagining and innovating at scale has been something Amazon has done for a long time and one of the reasons we’ve been the leader in technology infrastructure and data center invention, sustainability, and resilience. We're not done… there's still so much more to invent for customers.

The next era of datacenters is here. The demand for AI is growing rapidly, and with it comes the need to grow the cloud’s physical footprint. Historically, datacenters have been water-intensive and require using large amounts of higher carbon materials like steel. At Microsoft, we're building datacenters with sustainability in mind, and we're constantly innovating to find new ways to reduce our environmental impact. This includes: 🤝 A first-of-its-kind agreement with Stegra, backed by an investment from Microsoft’s Climate Innovation Fund (CIF) in 2024, to procure near zero-emissions steel from Stegra’s new plant in Boden, Sweden, for use in our datacenters. Powered by renewable energy and green hydrogen, Stegra's facility reduces CO2 emissions by up to 95% versus conventional steel production. By committing to purchase this green steel before it rolls off the line, Microsoft is sending a clear market signal, driving demand for cleaner materials and supporting Stegra’s growth. 💧 We also announced a major breakthrough to make our datacenters more sustainable: microfluidic in-chip cooling technology. Unlike traditional cold plates that sit atop chips, microfluidics brings cooling right inside the silicon itself. Engineers carve microscopic channels directly into the chip, letting liquid coolant flow through and absorb heat exactly where it’s generated. This approach is up to three times more effective than current methods. More efficient cooling allows datacenters to support powerful next-gen AI chips without ramping up energy use or investing in costly new gear. 💵 Through our CIF investments, we’ve catalyzed billions in follow-on capital for breakthrough solutions in low-carbon materials, sustainable fuels, carbon removal, and more. We just released a new whitepaper – Building Markets for Sustainable Growth – that distills five key lessons on how catalytic investment and partnership can move markets and accelerate a global transition in energy, waste, water, and ecosystems. Our journey toward sustainable datacenters is only beginning, and we recognize true progress requires collective action and investment. Read more from Building Markets for Sustainable Growth: https://msft.it/6041sq9xD

McKinsey & Company 𝗯𝗹𝘂𝗲𝗽𝗿𝗶𝗻𝘁 𝗳𝗼𝗿 𝗵𝗼𝘄 𝗯𝗮𝗻𝗸𝘀 𝗰𝗮𝗻 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗲𝘅𝘁𝗿𝗮𝗰𝘁 𝗿𝗲𝗮𝗹 𝘃𝗮𝗹𝘂𝗲 𝗳𝗿𝗼𝗺 𝗔𝗜: ⬇️ This is a full-stack, enterprise-grade architecture — built on agents, orchestration, and rewired workflows. The AI bank stack consists out of 4 key layers: ⬇️ 𝟭. 𝗘𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁 𝗟𝗮𝘆𝗲𝗿 This is the user layer — customers and employees. McKinsey calls for fully reimagined, intelligent, personalized experiences across all channels. → Multimodal chat (text, voice, image) → Omnichannel UX across mobile, contact center, branch → Digital twins for customer simulation and workforce training It’s all about a UI refresh and UX overhaul grounded in real AI. 𝟮. 𝗔𝗜-𝗣𝗼𝘄𝗲𝗿𝗲𝗱 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗠𝗮𝗸𝗶𝗻𝗴 This is the brain of the AI-first bank. And it’s not just predictive models anymore — it’s orchestrated agent ecosystems. → AI Orchestrators: Plan, reason, delegate across workflows → Domain Agents: Specialize in credit policy, fraud, risk, legal → Copilots: Embedded in workflows to guide users and automate decisions McKinsey reports 20–60% productivity gains in decision-making with this approach. 𝟯. 𝗖𝗼𝗿𝗲 𝗧𝗲𝗰𝗵 & 𝗗𝗮𝘁𝗮 The foundation layer most banks underestimate — until GenAI models stall in production. → Vector databases → LLM orchestration and FinOps → Search and retrieval engines → ML pipelines → Secure data architecture → API infrastructure The goal: make data accessible, tools reusable, and infra invisible to the business. Without this, nothing scales. 𝟰. 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝗠𝗼𝗱𝗲𝗹 This is where the transformation wins or fails. Without rewiring the org, the tech doesn’t matter. → AI control towers to track value and set guardrails → Cross-functional teams across business, tech, and AI → Platform operating model for speed and alignment → Enterprise-wide reuse of AI capabilities If you're building isolated projects without shared assets or central coordination, you’re not transforming — you’re experimenting. 𝗪𝗵𝗮𝘁 𝘁𝗵𝗶𝘀 𝗮𝗹𝗹 𝗮𝗱𝗱𝘀 𝘂𝗽 𝘁𝗼? The banks that win won’t be the ones with the most pilots. They’ll be the ones that industrialize agents, orchestration, and rewired workflows, with full-stack coordination. Full McKinsey article: https://lnkd.in/dPaJzVK4 𝗜 𝗲𝘅𝗽𝗹𝗼𝗿𝗲 𝘁𝗵𝗲𝘀𝗲 𝗱𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁𝘀 — 𝗮𝗻𝗱 𝘄𝗵𝗮𝘁 𝘁𝗵𝗲𝘆 𝗺𝗲𝗮𝗻 𝗳𝗼𝗿 𝗿𝗲𝗮𝗹-𝘄𝗼𝗿𝗹𝗱 𝘂𝘀𝗲 𝗰𝗮𝘀𝗲𝘀 — 𝗶𝗻 𝗺𝘆 𝘄𝗲𝗲𝗸𝗹𝘆 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿. 𝗬𝗼𝘂 𝗰𝗮𝗻 𝘀𝘂𝗯𝘀𝗰𝗿𝗶𝗯𝗲 𝗵𝗲𝗿𝗲 𝗳𝗼𝗿 𝗳𝗿𝗲𝗲: https://lnkd.in/dbf74Y9E

I was listening to a panel of Customer Success (CS) leaders recently, and wow—this function is in the middle of a massive transformation! The world has shifted from growth at all costs to real focus on usage: In the last couple of years, every B2B company has struggled with customer retention even more than customer acquisition. You want to drive churn down? Usage. You want to drive downgrades down? Usage. You want to drive upgrades up? Usage. Customer Success needs to drive usage but also make sure that the entire company is focused on usage. CS leaders need to be more like marketers: They can’t just react to problems; they need to actively engage customers, much like marketers do. Proactive, engaging experiences build loyalty, not just putting out fires. The goal? Make CS as compelling and essential as your best marketing campaign. CS leaders need to go from operating in silos to orchestrating the entire customer journeys: Disconnected teams create disconnected experiences. CS leaders are stepping into a new role: journey orchestrators. They’re aligning sales, marketing, and support to deliver a seamless, cohesive customer journey. It’s no longer enough to excel at your piece of the puzzle—CS must ensure the whole puzzle comes together. CS leaders cant just deliver results on heroics, they need excellence in CS systems. Relying on heroic individual efforts isn’t sustainable. CS needs the right systems, tools, and data to operate at scale. Real-time product insights aren’t a nice-to-have—they’re a must. Excellence in systems, not just effort, is what will drive success in the age of usage. CS leaders have a tough job. So help them help you. Whether it’s investing in tools, aligning teams, or driving a culture of customer-centricity, the better your CS function, the stronger your business. 

As organizations increasingly adopt hybrid-cloud architectures, understanding the right path and tools is crucial for professionals aiming to deliver resilient, scalable, and efficient applications. Here’s a Cloud Native roadmap breaking down the skills and tools to master across critical domains. Dive in and explore the ecosystem that powers modern applications! 🔴 𝟭. 𝗟𝗶𝗻𝘂𝘅 𝗙𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝘀   Linux remains at the heart of cloud-native systems. Get comfortable with terminal commands, bash scripting, and distributions like Ubuntu and Red Hat for a solid start. 🟢 𝟮. 𝗡𝗲𝘁𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝗘𝘀𝘀𝗲𝗻𝘁𝗶𝗮𝗹𝘀   Protocols like HTTP, SSL, and SSH form the backbone of connectivity. Tools like Wireshark are invaluable for monitoring and securing network traffic. 🔵 𝟯. 𝗖𝗹𝗼𝘂𝗱 𝗦𝗲𝗿𝘃𝗶𝗰𝗲𝘀   The cloud is non-negotiable! Whether AWS, Azure, or Google Cloud, understanding SaaS, PaaS, and IaaS is key to harnessing the cloud's potential. 🟣 𝟰. 𝗦𝗲𝗰𝘂𝗿𝗶𝘁𝘆   Security is foundational in cloud-native environments. Tools like Open Policy Agent and Prisma provide the framework for enforcing policies and securing applications. 🟡 𝟱. 𝗖𝗼𝗻𝘁𝗮𝗶𝗻𝗲𝗿𝘀 & 𝗢𝗿𝗰𝗵𝗲𝘀𝘁𝗿𝗮𝘁𝗶𝗼𝗻   Containers revolutionized app deployment! Master Docker, Kubernetes, and service meshes like Istio to orchestrate, scale, and manage applications seamlessly. 🟠 𝟲. 𝗜𝗻𝗳𝗿𝗮𝘀𝘁𝗿𝘂𝗰𝘁𝘂𝗿𝗲 𝗮𝘀 𝗖𝗼𝗱𝗲 (𝗜𝗮𝗖)   IaC tools like Terraform, Chef, and Puppet automate infrastructure, ensuring consistency and efficiency across deployments. IaC is a must for scalable cloud-native applications. 🟢 𝟳. 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆   With tools like Prometheus, Grafana, and Elastic Stack, observability gives you the visibility needed to monitor, troubleshoot, and optimize performance in real time. 🔵 𝟴. 𝗖𝗜/𝗖𝗗   Continuous Integration and Delivery streamline deployments. GitLab, Jenkins, and GitOps practices (Argo) enable rapid, reliable application delivery. This roadmap covers essential areas for cloud-native development, from Linux fundamentals to CI/CD and observability. But, the cloud-native landscape is vast and rapidly evolving! Did I miss any critical tools or concepts? Whether it's a tool you swear by or an emerging trend you're excited about, drop it in the comments! 👇

"Service reliability math that every engineer should know" I think it's useful for engineers to understand what uptime and reliability mean in practice. These numbers paint a good picture of what's involved :) Now while service reliability is often reduced to a simple percentage, the reality is far more nuanced than those decimal points suggest. First, not all downtime is created equal. A single 8-hour outage has dramatically different business implications than 480 one-minute outages, even though both sum to the same annual downtime. This distinction is particularly relevant when considering service level agreements (SLAs) and how they’re measured. The impact of downtime also varies significantly based on when it occurs. Five minutes of downtime during peak business hours might cost more than an hour of downtime during off-hours. This temporal aspect of reliability is often overlooked in simple percentage calculations. Each additional nine of reliability typically requires an order of magnitude more engineering effort and operational complexity. Moving from 99.9% to 99.99% isn’t just a matter of being "10 times more reliable" – it often requires fundamental architectural changes: At 99.9% (8h 45m downtime/year), you might get away with single-region deployment and basic failover At 99.99% (52m 35s), you’re typically looking at multi-region deployment, sophisticated health checking, and automated failover At 99.999% (5m 15s), you need redundancy at every layer, real-time monitoring, and likely some form of active-active deployment At 99.9999% (31s), you’re dealing with advanced techniques like chaos engineering, automated canary deployments, and sophisticated traffic management While understanding the basic math of service reliability is crucial, the real engineering challenge lies in understanding the context, trade-offs, and business implications of reliability decisions. The next time you see a reliability requirement, don’t just think about the percentage – think about the entire socio-technical system required to achieve and maintain that level of service. The numbers are simple. The engineering reality behind them is anything but. #softwareengineering #programming

AI security/securing the use of AI is going to kill me. I use Claude Code almost daily. It's a problem.... Here's what I have to change AGAIN this week. Security researcher Ari Marzuk disclosed 30+ vulnerabilities across AI coding tools. Cursor. GitHub Copilot. Windsurf. Claude Code. All of them. He called it IDEsaster. The attack chain includes prompt injection, hijacking LLM context, and auto-approved tool calls executing without permission. Then, legitimate IDE features are weaponized for data exfiltration and RCE. Your .env files. Your API keys. Your source code. Accessible through features you thought were safe. Most studies I read claim that around 85% of developers now use AI coding tools daily. Most have no idea their IDE treats its own features as inherently trusted. 𝗦𝗼... 𝗮𝗳𝘁𝗲𝗿 𝗿𝗲𝘃𝗶𝗲𝘄𝗶𝗻𝗴 𝗔𝗿𝗶'𝘀 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵, 𝗵𝗲𝗿𝗲'𝘀 𝗜 𝘄𝗶𝗹𝗹 𝗯𝗲 𝗱𝗼𝗶𝗻𝗴... Be warned: All this is SO much easier said than done! Audit every MCP server connection. Checked for tool poisoning vectors where legitimate tools might parse attacker-controlled input from GitHub PRs or web content. Removed servers I couldn't verify. Disabled auto-approve for file writes. The attack chains weaponize configuration files and project instructions like .claude/settings.json and CLAUDE.md. One malicious write to these files can alter agent behavior or achieve code execution without additional user interaction. Move all credentials to a secrets manager. No .gitignored .env files in agent-accessible directories. API keys live in 1Password CLI. Environment variables inject at runtime through a wrapper script the LLM never sees. Start running Claude Code in isolated containers. Mounted volumes limited to specific project directories. No access to ~/.ssh, ~/.aws, or ~/.config. If the agent gets compromised, blast radius stays contained. Enable all security warnings. Claude Code added explicit warnings for JSON schema exfiltration and settings file modifications. These exist because Anthropic knows the attack surface. Add pre-commit hooks for hidden characters. Prompt injections hide in pasted URLs, READMEs, and file names using invisible Unicode. Flag non-ASCII characters in any file the agent might ingest. The fix isn't to stop using AI coding tools. The fix is to stop trusting them implicitly. What controls do you have for AI tools with write access to your codebase? 👉 Follow for more AI and cybersecurity insights with the occasional rant #AISecurity #DevSecOps

Yesterday, the open internet got its first toll booth. Not from courts or Congress, but from Cloudflare. The company, whose infrastructure touches roughly 20% of all global internet traffic, announced a significant shift: All new customer domains will now block AI bots by default. That means crawlers like OpenAI's GPTBot, Anthropic's ClaudeBot, and Google's Extended bot can no longer freely help themselves to your content. Unless you, the website owner, explicitly allow it, they’re cut off. Cloudflare also announced a Pay‑Per‑Crawl marketplace, where publishers can set terms for access - turning content into a licensed input, not a free good. With one product update, Cloudflare rewrote the default terms of engagement between AI and the web. From open to closed. From assumed permission to enforced consent. From passive scraping to active negotiation. Cloudflare has forced the question that’s been dodged for years: If AI systems are built on the backs of human expression, who owns the value they create? ▪️ For 20+ years, the internet operated on a basic exchange: Publish freely → Get discovered via search → Monetize attention. But AI broke that contract. Users now ask Claude or ChatGPT for answers. The models reply using human-created content - with no credit, no clickthrough, and no compensation. When publishing stops being a path to discovery - and starts being a donation to model training - incentives collapse. What Cloudflare did was reintroduce friction. Not to break the web. But to rebalance it. ▪️ What makes this moment so interesting is that the solution isn’t coming from regulators. It’s not the result of a moral awakening or a philosophical reckoning. It’s a default setting in a SaaS dashboard. But because of Cloudflare’s scale, it might as well be policy. This is capitalism at its best. You don’t always need top-down regulation to enforce guardrails. Sometimes market incentives create their own form of governance. In a world where governments often lag behind technology, infrastructure becomes policy. ▪️ Cloudflare may not have intended to reshape digital economics, but they’ve done exactly that. By forcing AI firms to ask, license, or walk away, Cloudflare has built the bones of a content licensing infrastructure, without ever saying the word “copyright”. ▪️ Let’s be clear: Cloudflare didn’t do this because it’s noble. They did it because it’s profitable. Their customers - TIME, Condé Nast, Reddit, News Corp - are tired of being strip-mined for training data without so much as a backlink. And Cloudflare is in the perfect position to intervene. I won’t be surprised if Fastly, Akamai, maybe even AWS follow. For 30 years, we assumed the internet had no gatekeepers. But it did. We just didn’t notice them until they started saying no. The principle is planted: Scraping without consent is not neutral. It’s extraction. We could see a more cohesive consent layer for the web emerge - driven not by altruism, but by incentive alignment.