Google Cloud: cost and sovereign AI
Announcements from Google Cloud Next show partners and vendors focusing on secure, cost‑governed AI — from automated BigQuery cost optimisation to Databricks collaboration and NetApp’s sovereign‑cloud storage options. Equinix also unveiled an AI‑native network layer for adaptive infra management, underscoring a market shift from demo‑led AI to operational controls and sovereignty features. (morningstar.com)(databricks.com)(voicendata.com)(itwire.com)
Google Cloud Next announcements this week showed enterprise artificial intelligence selling points shifting from model demos to cost controls, private infrastructure and data residency. (cloud.google.com) (prnewswire.com) At the Las Vegas event, SELECT by DoiT said on April 16 it opened an early-access program for an automated Google BigQuery cost observability and optimization product, with general availability planned for the third quarter. The company said teams at Google Cloud Next 2026 could see demos at booth No. 1509. (prnewswire.com) (select.dev) Databricks said April 16 that its Google Cloud push at Next 2026 centers on Gemini integrations, Google Marketplace distribution and joint customer work aimed at “governed AI.” Google’s Databricks marketplace page describes the service as a secure, cost-effective data and artificial intelligence platform on Google Cloud. (databricks.com) (cloud.google.com) NetApp said April 16 it expanded its Google Cloud relationship with a four-year enterprise agreement to speed deployment of NetApp storage inside Google Distributed Cloud air-gapped, a sovereign cloud platform delivered by World Wide Technology. The companies said the setup targets private-cloud artificial intelligence workloads in regulated and sovereignty-sensitive environments. (netapp.com) (morningstar.com) Equinix made the same turn on April 15, launching Fabric Intelligence as an “AI-native operational layer” for network infrastructure that uses AI agents to deploy, manage and monitor enterprise connectivity for AI workloads. The company said the product is meant to replace more manual software-defined networking steps with automated operations. (equinix.com) (itwire.com) The common thread is that enterprise buyers are asking how to run artificial intelligence inside budgets, compliance rules and existing infrastructure, not just how to access larger models. Google used Next to pitch the same operational layer across its own stack, including new Data Cloud features and AI Hypercomputer infrastructure updates. (cloud.google.com 1) (cloud.google.com 2) BigQuery sits at the center of that spending problem because it is Google’s serverless analytics engine, where customers pay for data processing rather than fixed hardware. Tools that watch queries, flag waste and automate tuning are being marketed as a way to keep analytics and AI projects from producing surprise bills. (cloud.google.com) (prnewswire.com) Sovereign cloud addresses a different constraint: where data sits, who can access it and whether systems can run disconnected from the public internet. NetApp’s air-gapped Google Distributed Cloud arrangement is aimed at governments and other organizations that need cloud-style services without giving up isolation or residency controls. (netapp.com) (voicendata.com) Databricks and Google are pitching governance as the bridge between those two pressures. Their April 16 materials tie Gemini model access to Databricks’ data platform and Google Cloud infrastructure, framing the partnership around secured data platforms and controlled deployment rather than open-ended experimentation. (databricks.com) (prnewswire.com) By mid-April 2026, the companies clustered around Google Cloud Next were competing less on who had an AI feature and more on who could make AI auditable, predictable and local enough for large enterprises to sign off. (cloud.google.com) (equinix.com)
