The common thread in three recently logged incidents is not a shared vulnerability or a single gang, but a shift in where breaches begin. In each case the entry point was a cloud relationship a company had already authorized: employee data held by a manufacturing giant, IAM credentials tied to a big data platform's AWS account, and authentication secrets sitting inside a hosting provider. The perimeter being tested in 2026 is the trust layer between US technology companies and the cloud vendors they depend on.
A Pattern of Borrowed Access
The three stories differ in target and outcome, yet they describe the same mechanics. As BleepingComputer reported, Taiwan Semiconductor Manufacturing Company confirmed that personal employee data was leaked online after a threat actor put 6.9GB of information allegedly stolen from the company up for auction, with the breach linked to a cloud provider. Separately, BleepingComputer reported that a threat actor used compromised identity and access management (IAM) credentials for a Qubole big data platform's AWS account to access and exfiltrate data from multiple customers' AWS environments. And cloud hosting provider Railway, also per BleepingComputer, said hackers exploited vulnerabilities in its infrastructure to steal customer authentication secrets and generate approximately $740,000 in fraudulent AI usage, potentially exposing credentials for up to 100,000 workspaces.
Read together, these are not three stories about weak passwords or unpatched servers in the ordinary sense. They are stories about the keys organizations hand to cloud platforms so those platforms can work on their behalf. When an attacker obtains those keys, the breach is not a break-in at all; it is a legitimate-looking login, executed with the authority the victim already granted.
The Credential, Not the Code, Is the Target
The TSMC case illustrates how far downstream a cloud-linked compromise can travel. Employee personal data is among the most sensitive material a manufacturer holds, and its appearance for auction signals that the value of a breach now sits in the records rather than the infrastructure. The detail that the incident is linked to a cloud provider matters for US technology companies because it places the failure outside the company's own walls, in a relationship it may manage through contracts and access reviews rather than direct control.
The Qubole-linked AWS campaign sharpens the point. Here the compromised material was IAM credentials, the mechanism that determines what a cloud identity is allowed to do. Because those credentials belonged to a platform account, the attacker could reach multiple customers' AWS environments. That is a structural feature of multi-tenant cloud services: one trusted identity, many downstream environments. For US enterprises that route data through third-party platforms running on AWS, the security of their own configuration is only part of the exposure. The posture of the platform's account is part of it too.
Railway adds the financial dimension. Stolen customer authentication secrets were converted into approximately $740,000 in fraudulent AI usage, and the company said credentials for up to 100,000 workspaces may have been exposed. The loss here is not only data but consumption: an attacker who can authenticate as a customer can run up bills on that customer's account. The potential exposure of up to 100,000 workspaces also means the blast radius is measured in customers, each of whom may need to rotate secrets and audit activity.
