THE APEX TIMES
Baseten said its high-performance inference service for open AI models can now run within Microsoft’s Foundry environment for enterprises looking to combine performance with governance controls tied to their Azure arrangements.
Baseten, an AI infrastructure company focused on running large language and other machine-learning models, says it has extended its “inference” offering to work inside Microsoft Foundry. Inference is the step where an AI model processes new requests, such as generating text, translating, or answering questions, rather than being trained from scratch.
The announcement positions the integration as a way for enterprises to deploy open models, meaning machine-learning models whose weights and code are released under an open license, with performance and reliability from Baseten while using Microsoft’s enterprise controls in Foundry. Foundry is presented by Microsoft as an environment for building and managing AI solutions with security, governance, and management features designed for corporate use.
According to the post, the deployment is tied to customers’ existing Azure agreement. The companies’ framing suggests the arrangement is meant to reduce friction for organizations that already run workloads on Azure and want to bring open-model inference into a governed environment rather than operating an end-to-end stack themselves.
Baseten’s message centers on “high-performance inference,” which typically refers to throughput and latency improvements and the engineering needed to make inference usable at scale. The company also highlights “speed and reliability” as the value proposition, aimed at helping organizations run open models without sacrificing responsiveness for production workloads.
Microsoft’s role in the integration is described in terms of Foundry’s security and governance. That matters because inference services in production can expose sensitive data, require auditing, and must fit into internal policy controls. For enterprises, governance features are often a gating factor when moving from experimentation to deployment, especially when the underlying models are not proprietary.
In a broader sense, the move reflects a continuing shift in enterprise AI toward hybrid approaches. Many organizations want the flexibility of open models and the ability to select from multiple model families, while still requiring a managed environment for deployment controls, identity and access policies, and operational oversight.
The announcement did not detail specific technical benchmarks, supported model families, deployment architectures, pricing, or timelines beyond the fact that the integration is now available under the companies’ arrangement. It also did not provide additional information on how customers will configure Baseten within Foundry, what monitoring artifacts are included, or which compliance or security attestations apply.
For companies evaluating production AI, the next indicates to watch are the documentation that accompanies the integration, such as integration guides, supported use cases, and any published performance data. It will also be important to see whether Microsoft and Baseten offer more clarity on operational topics like observability, data handling, and lifecycle management once deployments are running at scale.
Microsoft’s chief executive commented on Nvidia leadership turmoil, praising Jensen Huang and emphasizing that decisions about who runs the company would ultimately rest with the Nvidia board of directors.
The broker raised its Palantir outlook on Nov. 16, increasing its target to $248 from $215 while maintaining its Buy stance.
The company reiterated that its push to expand open-source pathways in AI development remains on track, even as leadership shifts following Jensen Huang’s sudden exit.
In comments described as emphasizing discipline rather than broad spending, Nvidia’s interim CEO said the company will keep supporting the AI ecosystem while it weighs how much to invest in “neoclouds” and other customers.
A widely held semiconductor ETF rose nearly as much as Nvidia after the chipmaker posted results that traders interpreted as a beat. The episode highlights how a single mega-cap can move an index even when it represents a minority weight.
The investment bank reduced its valuation target for Broadcom to $425 from $475, arguing the stock still offers roughly mid-20% upside based on its updated outlook.
On Nvidia’s third-quarter earnings call, interim chief executive Colette Kress delivered a direct, motivational message aimed at steadier sentiment among investors as the company faces intensifying scrutiny around its AI business momentum.
Market coverage of NVIDIA’s quarterly results highlighted commentary during the earnings call and renewed focus on the company’s forward-looking guidance, with investors looking for outlines on the durability of demand for its accelerated computing platform.
Board chairman Harvey Jones said Nvidia’s board is not considering Jensen Huang’s reinstatement, while the company’s guidance and results trajectory helped lift the stock after regular trading. Jones declined further comment on Huang.
A new post-market snapshot suggests Amazon’s stake in the AI startup will be just under 15% after the offering, with an option that could raise exposure, while Google is expected to control nearly 14%, together totaling about 30%.
Comments
Community guidelines. Comments may be removed for violations.
Loading comments...