Come To Think of It...

If you committed an AWS key to a public repo in 2022, here is the bad news: it probably still works. Truffle Security re-verified 10,616 publicly exposed AWS keys that had surfaced over the previous four years — in git history, Docker images, CI logs, and public datasets — and 88% of them still authenticated as of August 10, 2026. The numbers get more interesting in the corporate subset. Of the live keys, 768 give full control of a company's AWS account: 526 are root keys, 242 are IAM users carrying AdministratorAccess. Root is the scary one, because a root key can't be scoped down. In Truffle's words, it is the account, including the ability to close it. The median leaked key in the study is five years old. Five years. And most of them were never rotated. Where the keys come from matters. Truffle's scanners found 431,875 public findings resolving to 64,024 unique verified keys across 50,654 accounts, and the single largest source was not a git repository. It was Huggin...

AI is moving from the era of "chatting" to the era of "doing." We are witnessing the rise of agentic computing—where AI models don't just answer questions but execute complex, multi-step workflows. This shift changes everything for infrastructure providers. It's no longer enough to simply throw massive GPU clusters at a problem; agents require high-density CPU compute and enormous memory bandwidth to manage state, coordinate tasks, and handle the data pipelines that fuel them. Microsoft’s recent expansion of Azure with AMD’s Helios AI platform and EPYC processors is a direct response to this specialized demand. For example, their new HDv2 VMs are specifically designed for massive agentic workload adoption, providing the high-density CPU compute needed for data preparation and reinforcement learning. As agents become more autonomous, they will require even more complex isolation—the "sandbox" problem. While major providers like AWS and Google Clou...

It’s easy to think of the Linux kernel as a monolithic, constantly scrutinized fortress. We assume that if a bug is old, it’s either been found or it’s harmless. But "Januscape" just proved that sometimes, the most dangerous vulnerabilities are the ones that have been sitting quietly in the basement for nearly two decades. Discovered by security researcher Hyunwoo Kim, this guest-to-host escape flaw (CVE-2026-53359) stems from a use-after-free weakness in the shadow MMU emulation of KVM/x86. For anyone running KVM-based virtual machines on Intel or AMD hardware, the implications are sobering. If an attacker gains root access within a guest VM—a common scenario in multi-tenant public clouds—they can break out, execute code as root on the host, and potentially compromise every other tenant sharing that server. It’s the ultimate nightmare for cloud providers and high-security environments alike. What makes Januscape particularly unsettling is its longevity. This wasn't a...

It is a special kind of unsettling to realize that a critical vulnerability has been quietly sitting in the Linux kernel for sixteen years. We tend to think of modern security as a constant arms race, a high-speed chase between attackers and defenders. But the 'Januscape' flaw (CVE-2026-53359) reminds us that sometimes, the most dangerous threats aren't the rapid-fire zero-days, but the architectural ghosts that just... stay there. The flaw is a classic use-after-free vulnerability hidden within the shadow MMU emulation of KVM/x86. For anyone running KVM-based virtualization—which, if you are in the public cloud, means you are almost certainly part of the blast radius—this is a nightmare scenario. A guest-to-host escape means that an attacker who manages to get root access inside a virtual machine can break out and execute code as root on the underlying host. In a multi-tenant environment, that doesn't just compromise one user; it compromises every single guest runnin...

For the last couple of years, the headline-grabbing obsession in AI has been about parameters, compute, and the sheer scale of the next frontier model. We’ve been conditioned to believe that the winner of the AI race is whoever can squeeze the most intelligence out of the next batch of H100s. But the latest moves from the heavyweights suggest that the actual bottleneck for enterprise adoption isn't the intelligence itself—it's the implementation. Microsoft just announced the formation of the "Microsoft Frontier Company," a massive $2.5 billion initiative designed to embed 6,000 engineers and industry experts directly into customer organizations. This isn't just a consulting arm; it's a tactical deployment force meant to handle the messy, real-world engineering required to actually run these models in production. AWS followed suit with a $1 billion commitment to a similar forward-deployed organization, and Anthropic and OpenAI have been moving in this direct...

Snowflake announced a $6 billion multi-year infrastructure commitment to AWS on Wednesday, the kind of number that used to belong exclusively in the Anthropic and OpenAI columns. The press release leans hard into "agentic AI adoption" — that's the marketing framing, sure — but the actual architecture being described tells a different story. Snowflake is committing the bulk of that $6B to Graviton compute. Not GPU instances. Not the flashy inference clusters everyone writes about. Custom ARM-based processors designed for price-performance, not peak throughput. This is the same chip family that Meta just signed a multibillion-dollar deal to deploy for its own agentic AI workloads, and it turns out the real battleground for enterprise cloud spending isn't model licensing or software platforms — it's who controls the silicon underneath. The context here is worth paying attention to. AWS's custom chip business is now generating over $20 billion a year and growing...

There is something almost poetic about the US government's premier cybersecurity agency — the one whose job is literally to plug holes in critical infrastructure — getting outsmarted by a contractor who treated a public GitHub repository like a digital junk drawer. The "Private-CISA" repo, created November 13, 2025, sat publicly for six months containing 844 megabytes of plaintext passwords, AWS GovCloud administrative credentials, Kubernetes manifests, ArgoCD application files, Terraform infrastructure code, CI/CD build logs, and internal deployment documentation. Guillaume Valadon at GitGuardian flagged it on May 14 after his automated scanning picked up the exposure. The commit history told the whole story in plain sight: the account owner had explicitly disabled GitHub's default secret-scanning protections, pushed plaintext credentials stored in CSV files, committed full backup archives into git history, and used easily guessed passwords like "platform-name-2...

AWS slipped something quietly radical into last week's announcements, and I almost missed it under the pile of Bedrock updates and MCP server GA notices. Amazon Bedrock AgentCore Payments — now in preview — gives AI agents the ability to autonomously discover, access, and pay for APIs, MCP servers, web content, and even other agents. Not request permission. Not flag a human for approval. Just transact. Built in partnership with Coinbase and Stripe, it uses the x402 protocol — a modern take on HTTP 402 Payment Required — to handle everything from wallet authentication to stablecoin micropayments, all within spending limits the developer sets at the session level. Coinbase's x402 Bazaar MCP server already exposes more than 10,000 paid endpoints an agent can browse and buy from. Your agent wakes up, decides it needs real-time weather data or a premium legal research API to finish a task, gets a 402 response, negotiates payment, and keeps going. No human in the loop. No credit card...

One of the more revealing AI stories this week is not a dazzling model demo. It is AWS quietly shipping the kind of features that only become necessary when a technology is escaping the lab and wandering into finance, governance, and internal politics. On April 9, AWS added Amazon Bedrock cost allocation by IAM user and role, which means companies can finally attribute model spend to specific teams, projects, and applications instead of staring at one big mysterious AI bill and pretending that counts as strategy. A few days later, AWS also put Agent Registry into preview through Bedrock AgentCore: a governed catalog for agents, tools, skills, MCP servers, and related resources, complete with approval workflows, audit trails, and search. That pairing is the interesting part. The industry keeps talking about AI agents as if the main challenge is making them more capable. In practice, the next corporate headache is much more ordinary: figuring out who built what, who is allowed to use it,...

The most interesting enterprise AI story this week is not another model with a benchmark chart trying to look like destiny. It is the fact that the big vendors are finally talking like operators, finance people, and the poor soul who has to explain the cloud bill later. Google’s GKE Inference Gateway work is about squeezing more useful work out of shared accelerator pools by routing real-time and async inference through the same infrastructure instead of keeping separate GPU islands for every mood swing in demand. AWS is attacking the same maturity problem from a different side. Agent Registry is basically an admission that enterprises are going to accumulate fleets of agents, tools, and MCP-connected services whether they plan it well or not, while IAM principal cost allocation for Bedrock says the quiet part out loud: AI usage now has to be tagged, grouped, and explained like any other serious line item. That is not the romance of AI. That is the bookkeeping of AI, and honestly it is...

I think one of the more honest AI infrastructure stories right now is that the glamorous part is over and the traffic engineering part has begun. Google Cloud’s recent writing on GKE Inference Gateway and its guidance on reaching the efficient frontier of LLM inference point to the same boring, important truth: once you try to run large models as a real service, the hard part is no longer just model quality. It is deciding which requests get accelerator time, how to preserve low latency for live traffic, and how to stop expensive GPUs from spending their days in a weird half-idle limbo because nobody trusted the scheduler. That is less cinematic than another benchmark chart, but it is much closer to where production AI starts charging rent. The useful signal here is that Google is describing inference in terms systems people already understand. The gateway story is about workload separation, queue discipline, and smarter routing between real-time and async jobs that share the same ac...

The interesting part of enterprise AI is no longer the model demo. It is the queue. Google Cloud’s recent GKE work keeps circling the same unglamorous truth: once you try to run LLMs as an actual service instead of a conference prop, the hard part is deciding what gets GPU time, when, and under which latency promises. In one post, Google describes an Inference Gateway that lets real-time and async workloads share the same accelerator pool, with live traffic taking priority while batch jobs quietly eat the leftover capacity. In another, it lays out the bigger picture more bluntly: inference is a tradeoff surface between latency, throughput, and cost, and most teams are still operating below the efficient frontier because their routing and caching are dumb. That sounds dry until you remember what the alternative looks like: expensive GPUs sitting half-idle because nobody wanted the political risk of letting a document-indexing job share space with a chatbot. That is why I think the real ...

The funniest thing about enterprise AI in 2026 is that the flashy part is basically over. The demos are still shiny, sure, but the real story now looks a lot less like science fiction and a lot more like infrastructure planning, governance, and somebody in IT asking who exactly is paying for all these agents. Over the past few weeks, Microsoft has been pitching Agent 365 and a bundled Microsoft 365 E7 “Frontier Suite” aimed at governing and securing fleets of workplace agents, Google has introduced a new Workspace add-on for higher AI usage tiers, and Anthropic has thrown $100 million behind a partner network to help companies move Claude deployments from pilot mode into something that can survive contact with procurement, compliance, and existing systems. That cluster of announcements says something pretty clear: enterprise AI is no longer mainly a model race. It is becoming a packaging, controls, and implementation race. The clever model still matters, obviously, but once every vendo...