Google announced a broad set of BigQuery and data-platform changes at Google Cloud Next ’23 on August 29–30, 2023. The aim was to connect analytics, data-lake access, machine learning and generative AI in a more unified workflow—not to introduce one all-in-one product or eliminate the services underneath it. Since then, Google has described BigQuery Studio as generally available, while the launch-era status of individual features varied.
What Google announced at Cloud Next ’23
The announcements were a coordinated product strategy: bring more of the work from data preparation through analytics and AI into a connected Google Cloud environment. BigQuery was positioned as the analytical center, with BigQuery Studio as a shared workspace and BigLake, Vertex AI, Dataplex and Looker providing related capabilities.
- BigQuery Studio: A workspace for SQL, Python, Spark and notebooks, intended to help data practitioners work with shared assets and collaborate.
- Vertex AI integration: Ways to invoke Google foundation models and other model-inference capabilities from BigQuery workflows, including workflows involving unstructured data.
- Open table formats: BigLake support for Hudi and Delta Lake, alongside performance work for Apache Iceberg, to improve access to data-lake tables.
- BigQuery Omni: Cross-cloud joins and materialized views intended to analyze data across clouds while reducing the need to copy all source data.
- Duet AI: An announced assistant for SQL, Python, metadata discovery and analytics tasks across BigQuery, Looker and Dataplex.
- Governance and privacy: Emphasis on lineage, data quality, metadata management and privacy-oriented collaboration such as data clean rooms.
Google’s Next ’23 announcement presented these as connected parts of a data-and-AI platform. The goal was to reduce handoffs among separate tools; the announcements did not mean that every service, permission system or billing model would disappear.
What BigQuery Studio was designed to change
Before a unified workspace, a team might write warehouse SQL in one tool, process data with Spark elsewhere, use a separate Python notebook, manage metadata in a catalog and build or deploy models in another service. BigQuery Studio was intended to bring more of these activities into one working environment. Google’s August 30, 2023 announcement described support for SQL, Python and Spark-oriented workflows, plus collaboration, version history and data-discovery and governance features.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →#1 Best Overall
That is a workspace-level unification, not a replacement for the underlying cloud architecture. Production Spark processing, machine learning, orchestration, identity and governance can still depend on distinct services, configuration and permissions. Teams should assess the integrations they actually need rather than assume one interface removes operational complexity.
How the BigQuery and AI connection was meant to work
Google’s approach was to let teams bring model inference closer to analytical data. The intended benefit was fewer export-and-reload steps when applying models to enterprise information. Potential tasks included classifying text, extracting entities, analyzing documents or images, translating content and generating embeddings, then combining results with structured business data.
BigQuery object tables provide a structured-record interface to unstructured objects stored in Cloud Storage. Google also announced that the BigQuery ML inference engine was generally available on August 25, 2023 for inference with custom, remote and pretrained models. That status applies to the inference engine announcement, not automatically to every AI-related feature introduced at Next ’23. See Google’s posts on Vertex AI foundation models in BigQuery and the BigQuery ML inference engine GA.
Calling a model from an analytical workflow does not make every AI task SQL-only, guarantee accurate outputs or remove processing elsewhere. Model availability, quotas, latency, quality and charges still matter. Sensitive data also needs appropriate access controls, retention rules, auditability and model-risk review. A model’s output should be validated before it becomes a business decision or trusted dataset.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Why open table formats matter
Hudi, Delta Lake and Iceberg are open table formats intended to make lake data usable across multiple processing engines. Google’s announcement emphasized Hudi and Delta Lake support through BigLake and improvements for Iceberg. The strategic benefit is potential interoperability for organizations using Spark, Databricks, other cloud environments or on-premises systems, without requiring every dataset to be rewritten into a single proprietary warehouse format.
Open format support improves portability, but it does not guarantee identical behavior across engines. Transaction support, metadata handling, partitioning, catalogs, performance and available features can differ. Google’s later BigQuery capabilities announcement continued the lakehouse direction, including managed Iceberg and catalog interoperability; those later developments should not be confused with the status of each feature in 2023.
What BigQuery Omni added for multi-cloud data
At Next ’23, Google highlighted cross-cloud joins and materialized views in BigQuery Omni. The idea was to query or analyze data across cloud environments without first copying every source into Google Cloud. This can matter when an organization has data distributed across Google Cloud, AWS or Azure, or when residency rules constrain where data is stored.
“Without moving data” should be read as reducing the need for bulk replication, not as a promise of zero network traffic, zero egress cost or zero processing outside the source cloud. Cross-cloud work can require network connectivity, permissions in multiple environments, supported regions and compatible formats. Remote access may also affect performance and make incident diagnosis more involved. Whether querying remotely or copying data is cheaper depends on workload, data movement, query patterns and network pricing.
Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteWhat Duet AI was meant to do—and what still needs review
Google announced Duet AI in BigQuery, Looker and Dataplex as an assistant that could help draft or complete SQL and Python, suggest corrections, find metadata using natural language and support conversational data exploration. The stated aim was to reduce routine work and make data assets easier to discover, not to certify generated code as correct.
Generated SQL can use the wrong table, misread a business definition, create an unintended many-to-many join or scan more data than necessary. Before generated code is used in production, teams should:
- Check that it uses the intended tables, fields, joins and business definitions.
- Run a dry run or other cost check and set query limits appropriate to the workload.
- Test results against known values and review access policies.
- Require human approval for production changes and recurring reports.
Governance remains an organizational responsibility
Lineage, profiling, quality checks, metadata discovery and clean-room capabilities can help teams understand and control how data is used. Google positioned Dataplex and BigQuery Studio as part of that governance story, including privacy-centric analysis through BigQuery data clean rooms and Ads Data Hub. These tools can support governance, but they do not create sound ownership, accurate definitions or correct permissions by themselves.
Organizations still need to design least-privilege IAM access, dataset and table policies, row- and column-level controls where appropriate, regionalization, sensitive-data classification, audit monitoring, retention rules and model-output review. A unified workspace may make controls easier to reach; it does not remove the need to configure and operate them.
Best Value
What matured after the 2023 announcement
BigQuery Studio was initially announced as a preview. In a later platform update, Google described it as generally available and positioned BigQuery as a unified, AI-ready analytics platform supporting SQL, Python, PySpark and natural-language workflows. That later status updates the story for Studio, but it does not make every feature announced in August 2023 generally available or establish the current availability of every model, region or integration. Google’s later overview is BigQuery is a unified, AI-ready data analytics platform.
Google’s subsequent messaging continued to emphasize multimodal analytics, multiple processing engines and multi-cloud data access. Its data analytics innovations overview and 2025 Google Data Cloud update provide that later context. Product names, licensing and feature availability can change; verify the specific capability and region before designing around it.
When BigQuery is a sensible fit
BigQuery is most compelling when its workflow advantages align with where an organization’s data and teams already are. It may be a strong fit when:
- The organization already relies on Google Cloud, Vertex AI, Looker or related Google services.
- SQL-based analytics and serverless operations are priorities.
- Teams want to connect warehouse analysis with Python, Spark, machine learning or model inference.
- Data includes both structured records and files such as documents or images.
- Open-format or cross-cloud access has clear value and the organization can manage the associated network and governance needs.
It may be a weaker fit when most data and compute are established in another cloud and remote-access economics are unfavorable; when a workload needs an operational database rather than an analytical warehouse; or when teams require highly predictable costs and struggle to manage usage-based query and inference charges. Deep investment in another lakehouse, catalog or orchestration ecosystem can also make consolidation less valuable.
Free tools Windows power users keep installed
One-click scans. No signup required.
Compare platforms using data location, open-format and catalog support, AI integration, cost model, workload performance, developer experience, security, operational complexity, portability and existing team skills. BigQuery’s query, storage, ingestion and capacity costs vary by region, edition and workload; Vertex AI inference and cross-cloud networking can add separate costs. Use Google’s live BigQuery pricing and pricing calculator for the intended configuration rather than relying on a static price.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
