AWS tries a telco tack; Teradata's new vector
Today on Product Saturday: AWS spruces up its Outposts server gear for wireless carriers, Teradata jumps on the vector database train, and the quote of the week.
Today on Product Saturday: AWS spruces up its Outposts server gear for wireless carriers, Teradata jumps on the vector database train, and the quote of the week.
Welcome to Runtime! Today on Product Saturday: AWS spruces up its Outposts server gear for wireless carriers, Teradata jumps on the vector database train, and the quote of the week.
(Was this email forwarded to you? Sign up here to get Runtime each week.)
Calling card: AWS Outposts was one of the biggest things the cloud provider launched way back in 2018, giving companies that still wanted or needed to hug their own servers a way to run AWS services in their own buildings. We haven't heard as much about it since, given that generative AI sucked all the oxygen out of the room a long time ago, but this week AWS launched a new version of Outposts designed for telecommunications companies.
The new gear "addresses telcos’ needs for low-latency, migration and modernization, data residency, and local data processing use cases," AWS said in a press release. Telcos are notoriously conservative about the infrastructure strategies they use to power their networks, and winning them over has been a big priority for cloud providers over the last several years.
Security blanket: It's always a little stunning to learn how many enterprises have a fuzzy picture — at best — of exactly what is running on their infrastructure. That problem has likely only gotten worse amid the rush to deploy generative AI applications, and Google Cloud unveiled a new security service this week called AI Protection that it thinks might help.
"Effective AI risk management begins with a comprehensive understanding of where and how AI is used within your environment," the company said in a blog post. AI Protection allows customers to identify how sensitive data is being used by AI apps, set controls and guardrails based on that information, and detect and respond to incoming threats targeting that data.
Space and time: It's clear that vector storage is here to stay as enterprises build generative AI apps, but it's less clear which companies will emerge as the winner of this exploding market. Teradata threw its hat in the ring this week with the launch of Enterprise Vector Data Store, an "in-database solution" designed for hybrid cloud customers.
"Enterprise Vector Data Store is designed to be a performant way to enable use cases that require vector capabilities and RAG applications," the company said in a release. “Almost all data platform providers are adding support for the storage and processing of vectors as they update their products to support enterprise requirements for applications based on generative AI,” Matt Aslett of ISG told InfoWorld.
Business ready: On the other side of the vector database market you'll find the purpose-built upstarts, such as Pinecone and Qdrant. All enterprise tech startups eventually learn that there are a few must-have features in order to be taken seriously by big business, and this week Qdrant added several of those to its Qdrant Cloud managed vector database.
"The latest updates include single sign-on (SSO), cloud role-based access control (RBAC), granular database API keys for granular RBAC, advanced monitoring and observability with Prometheus/OpenMetrics to connect external monitoring systems, and a cloud API for seamless automation," Qdrant said in a press release. It's far from clear whether enterprises want to put a startup at the heart of their generative AI tech stack, but a substantial portion of them won't even have the conversation without features like the ones Qdrant rolled out this week.
Hop to it: A lot of companies that really wanted to ship generative AI apps last year ran into the last-mile problem, where they couldn't get their proof-of-concept apps into production due to concerns about accuracy and performance. This week JFrog introduced JFrog ML, a new addition to its CI/CD software platform that it thinks can make those deployments easier by getting data scientists, developers, and ops engineers on the same page.
"By treating ML models as software packages from the start of development and converging ML model management and software development into a single source of truth, the friction and errors between stages and teams can be significantly reduced," the company said in a press release. It's another argument that AI agents are just software, not "digital labor platforms" that companies like Salesforce and Workday have promoted.
If you're using Kubernetes you probably already know you're spending too much money provisioning compute resources, but CastAI has the receipts, as seen on CIO Dive. According to a survey of 2,100 companies running Kubernetes clusters on the Big Three cloud providers, "almost every cluster was underutilized, with organizations using an average of just 10% of cloud CPU capacity and less than one-quarter of memory capacity provisioned over the 12-month period."
“Elon is really high intensity—high expectations. But it comes at the expense of your sanity.” — Lionel Branscomb, a contractor who is helping build Project Stargate with data-center startup Crusoe in Texas, describing the shadow President of the United States to The Information.
Phil Venables, the first CISO in Google Cloud's history, announced Friday that he is leaving the company to pursue a new, unspecified venture.
Anysphere, which developed the increasingly popular Cursor code editor, has hit $100 million in annual recurring revenue and is in talks to raise a financing round that would value the company at $10 billion, according to Bloomberg.
Thanks for reading — see you Tuesday!