Securing AI on your own terms: Anaconda acquires Enkrypt AI

Every enterprise can tell you how fast its AI agents are moving, but fewer can tell you what those agents are doing. Anaconda has acquired Enkrypt AI, the security and safety layer enterprises use to test, monitor, and govern their full AI agent stack.

In the two months before the announcement, Enkrypt AI scanned more than 268,000 agent tools across 25,000 MCP servers and found more than 143,000 vulnerabilities affecting 73% of those servers. Enkrypt brings pre-deployment red-teaming across 300+ attack categories, runtime guardrails that deploy inside your own environment, and compliance automation for the NIST AI Risk Management Framework and the EU AI Act.

Gartner® names Kilo Code in Coolest Vendor Innovations in AI Coding Agents

Orchestrating, supervising, and governing agent behavior help vendors stand out. Gartner evaluates four vendors in this report, including Kilo Code, now part of Anaconda, and lays out the five value paths shaping AI-native software engineering through 2030.

As Gartner puts it, “the coolest innovations in this emerging market are capabilities that enable developers to orchestrate, supervise and govern AI agent behavior.” That is the gap Anaconda acquired Kilo Code to close: open, model-agnostic coding agents running inside governed environments.

Only 17% of IT leaders have a unified AI observability strategy

When a model produces a decision you have to explain, the answer lives in the environment, dependencies, and data behind it. Omdia surveyed 500 IT leaders across North America and the UK, and 17% reported a unified observability strategy spanning teams, infrastructure, and governance.

The report maps where coverage drops off across the AI lifecycle, what buyers weigh when selecting a partner, and which integration costs they rank above license price. Commissioned by Anaconda and distributed under license from Informa TechTarget, Inc.

main-x is generally available, with CVE data and SBOMs

main-x is an authenticated channel that extends main with more than 13,000 additional packages, built by Anaconda with license validation, malware scanning, typosquat protection, and compatibility testing. Log in, add one channel, and thousands more packages are at your fingertips.

General availability adds what security and compliance teams ask for before they approve a package: CVE data and SBOMs, available in Anaconda Platform. New packages land every week.

NVIDIA is buying Hugging Face for $12.9 billion

NVIDIA has agreed to acquire Hugging Face for $12.9 billion, according to The Information. It would rank among NVIDIA’s largest acquisitions to date, and it places the hub where developers publish and download open-source models under the ownership of the company making most of the hardware those models run on. Reporting indicates the agreement is not yet signed.

Three deals in four weeks: Qualcomm closed its acquisition of Modular on July 29, Stripe agreed to buy OpenRouter on August 19, and now this. The infrastructure layer under open source AI is consolidating, which makes knowing where your models and packages come from even more crucial.

Falling per-token prices are not lowering your AI bill

Per-unit inference costs keep dropping while total enterprise AI spending climbs. AIwire walks through why: cheaper tokens make agentic workloads viable that were not viable before, so consumption grows faster than price falls.

The reported figure is that 83% of organizations say they need to upgrade their infrastructure before they can fully support agentic AI. Costs you cannot attribute to a specific workload are an observability problem before becoming a budget problem.