Log entry Nº 642 · the stack watch
Bearing E 090° · November 18, 2026 · 3 min read
E 090°Watch log
ThoughtSpot Launches Customer Hosted for On-Premises, Sovereign AI Analytics
ThoughtSpot's Customer Hosted brings on-premises and private AWS/Google Cloud deployment, private AI models and in-VPC control to sovereign-data buyers.
By Tom Whitfield3 min read650 words

ThoughtSpot on Wednesday launched Customer Hosted, a deployment option that lets enterprises run its data and AI analytics platform on premises and in private deployments on AWS and Google Cloud. The release answers a market spending $80.4 billion on sovereign clouds this year, up 35%, and forecast by Gartner to reach $110.6 billion in 2027.
The move targets buyers in regulated industries where data sovereignty — the principle that data falls under the laws of the country or region where it is collected and stored — has become a gating factor for AI adoption. AI agents now call on data directly, and controlling every action an agent takes, including how it accesses and moves data between systems, is far harder than overseeing human behavior. Stricter AI regulations, especially in Europe, raise the stakes.
The cost of getting this wrong is concrete. Meta has paid more than $1.5 billion in fines across Texas and Ireland over data capture and mishandled passwords. LinkedIn paid $336 million for violating the EU's General Data Protection Regulation. Uber was fined $324 million in the Netherlands for inadequate storage and transfer safeguards for its European data.
What does Customer Hosted change for practitioners?
The option delivers the same capabilities as ThoughtSpot's cloud-native platform, but inside the customer's own security perimeter. Specific features include:
- Support for private AI models, so customers do not have to use public large language models such as Anthropic's Claude or OpenAI's ChatGPT
- Flexible network integration to connect Customer Hosted with internal systems and operationalize data for AI and analytics
- Governance and log access through virtual private cloud service controls
- A pricing structure unique to each deployment that lets existing cloud customers apply part of their cloud license cost to the new version
Devin Pratt, an analyst at IDC, pointed to one design choice as the differentiator: Customer Hosted places the control plane in the customer's AWS or GCP environment rather than in ThoughtSpot's. "Control only counts when it spans every layer," he told TechTarget. "Relocating the control plane into the customer's cloud turns a security assurance into something a team can verify."
"Customers shouldn't have to choose between a modern agentic analytics platform and full control of their infrastructure," ThoughtSpot CEO Ketan Karkhanis told TechTarget. "[Customers] want agentic analytics, but they also need to control where their platform runs and how it is managed. As AI moves from working with metadata to reasoning over business data, those questions get more pressing."
How does it compare with rivals?
Self-managed deployments are not new. Cloudera, Tableau, Qlik and Strategy already offer on-premises and private cloud data and AI deployments. William McKnight, president of McKnight Consulting, said the launch is still meaningful: "It [isn't] novel against [competitors] which have offered self-managed deployments for a while, but it does bring them to parity allowing highly regulated enterprises to deploy modern, natural-language search and sovereign agentic AI entirely within their own cloud security perimeter."
Pratt drew the same line: "Self-hosted BI is established, [but] self-hosted agentic analytics, with private models, inside the customer's own boundary, is not."
For vendors targeting heavily regulated sectors, McKnight called such deployment options "a commercial necessity, not a nice-to-have."
What comes next?
McKnight said ThoughtSpot should now add first-party evaluation capabilities to help customers detect and prevent agent drift over time, and possibly expand AgentSpot into ThoughtSpot Embedded so software vendors can natively deliver collaborative agentic workflows in their own applications.
Karkhanis, without naming specific capabilities, said the roadmap focuses on agentic analytics at every decision point, embedded AI and BI inside organizational applications, and delivering intelligence to agents through AgentSpot.
Pratt's recommendation: add Microsoft Azure support. "A large share of the buyers who need this are on Azure," he said. "Meeting them there is the clearest next step."
Plot points
- ThoughtSpot launched Customer Hosted on Wednesday for on-premises and private AWS/Google Cloud deployment
- Gartner pegs sovereign cloud spend at $80.4 billion this year, up 35%, rising to $110.6 billion by 2027
- Customer Hosted supports private AI models and puts the control plane in the customer's own cloud
- Existing cloud customers can apply part of their license cost to the new deployment model
- Competitors Tableau, Qlik, Cloudera and Strategy already offer self-managed deployments
via techtarget.com (Source)
More from Tom Whitfield
Show full bio
News editor covering money, funding and market shifts around marketing technology at Martech Beacon.
38 articles
Neighbouring fixes
- AI Assistants' Rogue Acts Are Chilling Consumer Adoption
- Only 10.8% of AI Agent Initiatives Reach Full Scale as Budgets Surge
- Gartner: Databricks CustomerLake Erodes Stand-Alone CDPs
- OpenAI Draws a Hard Line on ChatGPT Data as It Builds Its Ads Business
- Alteryx launches VURA governance suite to make AI show its receipts