BigQuery is 5x larger than Snowflake or Databricks. What Google is doing for it to be even bigger

Google Cloud has announced a number of new features at its Google Cloud Nextlast week with at least 229 announcements.

In the midst of all that news, including new AI chips, agentic AI capabilities andupdates to databasesGoogle Cloud made some major moves with its BigQuery service. BigQuery Unified Governance is one of the new capabilities that helps organizations discover, trust and understand their data assets. The governance tools are designed to help overcome key AI adoption barriers by ensuring data accessibility, quality and trustworthiness.

Google is facing a huge challenge as it competes with rivals in the enterprise-data space.

BigQuery, which has been available since 2011, has seen a significant increase in its capabilities and users in the last few years. BigQuery appears to be a major business for Google Cloud. Google Cloud Next revealed for the very first time how big this business is. Google reported that BigQuery had five-times the number of clients of Snowflake and Databricks.

This is the first time we’ve had permission to post a customer statistic, which was wonderful for me,” Yasmeen Ahmed, managing director of Google Cloud data analytics, told VentureBeat. “Databricks, Snowflake are the only other enterprise data warehouse platforms on the market. We have five times as many customers as either of them.”

How Google is improving BigQuery in order to advance enterprise adoption.

Google claims that it has a larger user base than its competitors, but it’s also not taking the pedal off. Google Cloud Next and other hyperscalers have announced new capabilities in recent months to help advance enterprise adoption.

Access to the right data that meets service level agreements is a key challenge for enterprise AI. Google cites Gartner research that shows organizations who do not have AI-ready data practices will abandon over 60% of AI projects if they fail to meet business SLAs. This challenge is a result of three persistent problems that plague enterprise management:

Fragmented data-silos

Google’s BigQuery Unified Governance represents a significant departure in traditional approaches, as it embeds governance capabilities directly into the BigQuery platform instead of requiring separate tools and processes.

BigQuery unified Governance: A technical deep-dive

The new BigQuery universal catalogue is at the heart of Google’s announcement. It powers the BigQuery unified government. The universal catalog, unlike traditional catalogs which only contain basic table/column information, integrates three distinct metadata types:

1. Technical/Physical metadata: Schema Definitions, Data Types and Profiling Statistics.
2. Metadata for business: Business glossary, descriptions and semantic context.
3. Runtime Metadata: Query patterns and usage statistics for technologies such as Apache Iceberg.

With this unified approach, BigQuery can maintain a comprehensive view of data assets throughout the enterprise. Google’s integration of Gemini, their advanced AI model directly into the governance layer via what they call the Knowledge Engine, makes the system especially powerful.

By discovering relationships between datasets and enriching metadata with context from the business, the knowledge engine enhances governance.

Key features include semantic search using natural language understanding, automated meta data generation, AI-powered relationships discovery, data products related to packaging assets, a glossary for business, automatic cataloging both structured and nonstructured data, and automated anomaly detection.

Google’s AI strategy is more important than benchmarks.

Ahmad said, “I think the industry is too focused on the individual leaderboards and Google is actually thinking holistically about this problem.”

The comprehensive approach covers the entire enterprise data-lifecycle and answers critical questions, such as: How can you deliver on trust in your data? How do you deliver scale? How do you deliver governance and security?

By combining innovations at each layer and innovating together, Google has created a real-time flywheel for data activation, where metadata, lineage, and quality are generated instantly as soon as the data is captured. This is true regardless of its format, type, or location.

Models are important. Ahmad explained that the introduction of thinking models such as Gemini 2.0 has opened up Google’s data platform.

She said that a year ago, if you asked GenAI to answer an important business question, you had to break it into several steps. “Suddenly, the thinking model can come up a plan… You don’t have to hard code it a way to build a strategy.” It knows how plans are built.”

She said that you can now easily have a data engineer agent build a pipeline of three steps or ten steps. The integration of Google’s AI capabilities with enterprise data has revolutionized what is possible.

Real-world impact of the Enterprise: How it benefits

Levi Strauss & Company (19459044) is a compelling example that unified data governance transforms business operations. The 172-year old company is using Google data governance capabilities to transition from a wholesale business into a direct-to consumer brand. Vinay Narayana – who is responsible for data and AI platform engineering at Levi’s – presented his organization’s case at Google Cloud Next.

Narayana stated, “We aim to empower our analysts with accurate real-time data.” “Before embarking on our journey to create a new platform, I discovered that users faced a variety of challenges. Our business users did not know where data resided, and even if they did, they had no idea who owned the data. If they got access, no documentation was available.”

Levi’s created a data platform in Google Cloud to organize data products by business domains, making them discoverable via Analytics Hub (Google’s marketplace for data). Each data product comes with detailed documentation, lineage and quality metrics.

Results have been impressive. “We are 50x quicker than our legacy data platforms, and that is at the low end.” Narayana stated that a significant number of visualizations were 100x faster. “We have 700 users using the platform daily.”

Verizon is also using Google’s governance tool as part of One Verizon Data to unify data that was previously siloed across business units.

Arvind Rajagopalan is the AVP of Data Engineering, Architecture and Products at Verizon. He said, “This will be the largest telco warehouse in North America using BigQuery.” Verizon said during a Google Cloud Next presentation.

This company’s data estate consists of 3,500 users, who run 50 million queries per year, 35,000 data pipes, and more than 40 petabytes.

Ahmad provided a number of other user examples in a spotlight session during Google Cloud Next. Radisson Hotel Group customized their advertising at large scale by training Gemini models using BigQuery data. Teams saw a 50% increase of productivity, and revenue from AI-powered campaign rose by over 20%. Gordon Food Service migrated from SQL Server to BigQuery. This ensured their data was AI-ready and increased adoption of customer facing apps by 96%.

What’s the “big” difference? Exploring the competitive environment

In the enterprise data warehouse market, there are many vendors, including Databricks and Snowflake. Microsoft has Synapse, while Amazon has Redshift. In recent years, all of these vendors have developed various AI integrations.

Databricks acquired Mosaic for $1.3 billion and has expanded their own AI capabilities. Amazon Redshift will support generative AI by 2023. AmazonQ will help users create queries and get better answers. Snowflake, on the other hand, has been developing tools and partnering with large language models (LLMs)including Anthropic.

When asked about comparisons to Microsoft’s offerings specifically, Ahmad argued Synapse was not an enterprise data platforms for the types of use-cases that customers use BigQuery. She said, “I believe we’ve jumped ahead of the industry because we’ve worked with all the pieces.” “We have the best model. It’s integrated in a data-stack that understands how agents operate.”

Using this integration, BigQuery has adopted AI capabilities rapidly. According to Google, the use of Google’s AI in BigQuery by customers for multimodal analyses has increased 16 times over the past year.

What this means for enterprises adopting AI.

For companies already struggling with AI implementations, Google’s integrated governance approach may offer a more streamlined route to success than putting together separate data management systems and AI systems.

Ahmad’s claim that Google “leapfrogged’ competitors in this area will be scrutinized as organizations put these capabilities to work. The customer examples and technical details show that Google has made significant strides in addressing enterprise AI adoption’s most challenging aspect.

The key question for technical decision-makers who evaluate data platforms is whether this integrated approach provides enough additional value to justify moving from existing investments in specialized platform, such as Snowflake and Databricks. And whether Google can continue its current innovation pace while competitors respond.