We introduced a number of new improvements with our autonomous knowledge to AI platform powered by BigQuery, alongside our unified, trusted, and conversational Looker BI platform:
125. BigQuery pipelines, now GA, helps constructing knowledge pipelines.
126. BigQuery knowledge preparation, now GA, transforms and enriches knowledge.
127. BigQuery anomaly detection, now in preview, maintains knowledge high quality and automates metadata era.
128. Information science agent, now GA, is embedded inside Google’s Colab pocket book, offers clever mannequin choice, enabling scalable coaching, and quicker iteration.
129. Looker conversational analytics, in preview, lets enterprise customers work together with knowledge utilizing pure language.
130. The Looker Conversational Analytics API, now in preview, lets builders construct and embed conversational analytics into functions and workflows. Enroll to achieve entry.
131. BigQuery data engine, in preview, leverages Gemini to investigate schema relationships, desk descriptions, and question histories to generate metadata on the fly, mannequin knowledge relationships, and advocate enterprise glossary phrases.
132. BigQuery semantic search, is now GA, offering AI-powered knowledge insights and throughout BigQuery, grounding AI and brokers in enterprise context.
133. BigQuery’s contribution evaluation function, now GA, helps you pinpoint the important thing components (or combos of things) answerable for essentially the most vital adjustments in a metric.
134. We added Gemini in BigQuery options into present BigQuery pricing fashions throughout all BigQuery compute pricing choices. Categorical your curiosity in attempting out the brand new options.
135. BigQuery pipe syntax is GA, letting you apply operators in any order and as typically as you want, and is suitable with most traditional SQL operators.
Then, for knowledge science and analyst groups, we added AI-driven knowledge science and workflows as a part of BigQuery pocket book:
136. New clever SQL cells perceive your knowledge’s context and supply sensible recommendations as you write code, and allow you to be part of knowledge sources straight inside your pocket book.
137. Native exploratory evaluation and visualization capabilities in BigQuery make it straightforward to discover knowledge, in addition to add options to allow simpler collaboration with colleagues. Information scientists may also schedule analyses to run and refresh insights periodically.
138. The brand new BigQuery AI question engine lets knowledge scientists course of structured and unstructured knowledge along with added real-world context, co-processing conventional SQL alongside Gemini to inject runtime entry to real-world data, linguistic understanding, and reasoning skills.
139. Google Cloud for Apache Kafka, now GA, facilitates real-time knowledge pipelines for occasion sourcing, mannequin scoring, messaging and real-time analytics.
140. Apache Spark workloads inside BigQuery, now in preview, execute serverlessly, that’s, on a totally managed platform.
141. New dataset-level insights in BigQuery knowledge canvas, in preview, floor hidden relationships between tables and generate cross-table queries by integrating question utilization evaluation and metadata.
142. BigQuery ML contains the brand new AI.GENERATE_TABLE in preview to seize the output of LLM inference inside SQL clauses.
143. BigQuery ML now helps open-source and Anthropic’s Claude, Llama, and Mistral fashions, hosted on Vertex AI.
144. BigQuery vector search features a new index sort, now GA, based mostly on Google’s ScaNN mannequin that’s coupled with a CPU-optimized distance computation algorithm for scalable, quicker and extra cost-efficient processing.
145. The preview of BigQuery ML’s pre-trained TimesFM mannequin developed by Google Analysis simplifies time-series forecasting.
146. We built-in new Google Maps Platform datasets straight into BigQuery, to make it simpler for knowledge analysts and determination makers to entry insights.
147. As well as, Earth Engine in BigQuery brings one of the best of Earth Engine’s geospatial raster knowledge analytics straight into BigQuery. Be taught extra right here.
148. GrowthLoop launched its Compound Advertising Engine constructed on BigQuery with Development Brokers powered by Gemini, so advertising and marketing can construct customized audiences and journeys that drive quickly compounding development.
149. Informatica expanded its providers on Google Cloud to allow refined analytical and AI governance use circumstances.
150. Fivetran launched its Managed Information Lake Service for Cloud Storage with native integration with BigQuery metastore and automated knowledge conversion to open desk codecs like Apache Iceberg and Delta Lake
151. DBT is now built-in with BigQuery DataFrames and DBT Cloud is now on Google Cloud.
152. Datadog launched expanded monitoring capabilities for BigQuery, offering granular visibility into question efficiency, utilization attribution, and knowledge high quality metrics.
BigQuery’s autonomous knowledge basis offers governance, orchestration for numerous knowledge workloads, and a dedication to flexibility through open codecs. Bulletins on this space embrace:
153. BigQuery makes unstructured knowledge a first-class citizen with multimodal tables in preview, bringing wealthy, complicated knowledge sorts alongside structured knowledge for unified storage and querying through the new ObjectRef knowledge sort.
154. BigQuery governance in preview offers a single, unified view for knowledge stewards and professionals to deal with discovery, classification, curation, high quality, utilization, and sharing.
155. The brand new BigQuery common catalog brings collectively an information catalog (previously often known as Dataplex Catalog) and a totally managed, serverless metastore, now typically obtainable.
156. BigQuery metastore, now GA, allow engine interoperability throughout BigQuery, Apache Spark, and Apache Flink engines, with assist for the Iceberg Catalog.
157. BigQuery enterprise glossary, now GA, helps you to outline and administer firm phrases, establish knowledge stewards for these phrases, and fix them to knowledge asset fields.
158. BigQuery steady queries, now GA, allow on the spot evaluation and actions on streaming knowledge utilizing SQL, no matter its authentic format.
159. BigQuery tables for Apache Iceberg in preview, helps you to join your Iceberg knowledge to SQL, Spark, AI and third-party engines.
160. New superior workload administration capabilities, now GA, scale assets, handle workloads, and assist guarantee their cost-effectiveness.
161. BigQuery spend commit, now GA, simplifies buying, unifying spend throughout BigQuery knowledge processing engines, streaming, governance, and extra.
162. BigQuery DataFrames now has AI code help capabilities in preview, letting you employ pure language prompts to generate or counsel code in SQL or Python, or to clarify an present SQL question.
163. SQL translation help, now GA, is an AI-based translator that allows you to create Gemini-enhanced guidelines to customise your SQL translations, to speed up BigQuery migrations.
164. Catalog metadata export, GA, allows bulk extract of catalog entries into Cloud Storage.
165. BigQuery can now carry out automated at-scale cataloging of BigLake and object tables, now GA.
166. BigQuery managed catastrophe restoration is now GA, that includes automated failover coordination, steady near-real-time knowledge replication to a secondary area, and quick, clear restoration throughout outages.
167. New workload administration capabilities in preview embrace reservation-level truthful sharing of slots, predictability in efficiency of reservations, and enhanced observability via reservation attribution in billing.
Looker, is including a number of recent conversational and visible capabilities, geared toward making BI accessible and helpful to all customers, accelerated by AI.
168. Gemini in Looker options are actually obtainable to all Looker platform customers, together with Conversational Analytics, Visualization Assistant, System Assistant, Automated Slide Era, and LookML Code Assistant.
169. Code Interpreter for Conversational Analytics is in preview, permitting enterprise customers to carry out forecasting and anomaly detection utilizing pure language with no need deep Python experience. Be taught extra and join it right here.
170. New Looker stories function an intuitive drag-and-drop interface, granular design controls, a wealthy library of visualizations and templates, and real-time collaboration capabilities, now within the core Looker platform.
171. With Google Cloud’s acquisition of Spectacles.dev, builders can automate testing and validation of SQL and LookML adjustments utilizing CI/CD practices.
Firebase
172. The brand new Firebase Studio, obtainable to everybody in preview, is a cloud-based, agentic growth surroundings powered by Gemini that features every part builders must create and publish production-quality full-stack AI apps rapidly, multi function place. Gemini Code Help brokers can be found through personal preview.
173. Genkit, an open-souce framework for constructing AI-powered functions, utilizing your most popular language, now has early assist for Python and expanded assist for Go. Attempt this template in Firebase Studio to construct with Genkit.
174. Vertex AI in Firebase now contains assist for the Stay API for Gemini fashions, enabling extra conversational interactions in apps resembling permitting clients to ask audio questions and get responses.
175. Firebase Information Join is now GA, providing the reliability of Cloud SQL for PostgreSQL with on the spot GraphQL APIs and type-safe SDKs.
176. Firebase App Internet hosting can also be GA, offering an opinionated, git-centric internet hosting resolution for contemporary, full-stack net apps.
177. A brand new App Testing agent inside Firebase App Distribution, additionally in preview, prepares cell apps for manufacturing by producing, managing, and executing end-to-end checks.
Google Cloud Consulting
Google Cloud Consulting launched a number of new pre-packaged service choices:
178. Agentspace Accelerator offers a structured strategy to connecting and deploying AI-powered search inside organizations, so workers can simply acquire entry to related inner info and assets once they want it.
179. Optimize with TPUs helps clients migrate workloads to our purpose-built AI chips, TPUs.
180. Oracle on Google Cloud lets clients mix Oracle databases and functions with Google Cloud’s superior platform and AI capabilities for enhanced database and community efficiency.
181. We expanded entry to Supply Navigator, a collection of confirmed supply methodologies and greatest practices to assist with migrations and know-how implementations to clients in addition to companions, in preview.
Networking
Google’s international community is the spine of Google Cloud’s providers, and depends on an extended listing of improvements.
182. Cloud WAN, a Cross-Cloud Community resolution, is a totally managed, dependable, and safe enterprise spine that makes Google’s international personal community obtainable to all Google Cloud clients. Cloud WAN delivers as much as 40% improved community efficiency, whereas decreasing whole value of possession by as much as 40%. Learn extra right here.
183. The brand new 400G Cloud Interconnect and Cross-Cloud Interconnect, obtainable later this yr, provides as much as 4x extra bandwidth than our 100G Cloud Interconnect and Cross-Cloud Interconnect, offering connectivity from on-premises or different cloud environments to Google Cloud.
184. Construct large AI providers with networking assist for as much as 30,000 GPUs per cluster in a non-blocking configuration, obtainable in preview now.
185. Zero-Belief RDMA safety helps you safe your high-performance GPU and TPU visitors with our RDMA firewall, that includes dynamic enforcement insurance policies. Obtainable later this yr.
186. Get accelerated GPU-to-GPU communication, with as much as 3.2Tbps of non-blocking GPU-to-GPU bandwidth with our high-throughput, low-latency RDMA networking, now typically obtainable.
187. Leverage GKE Inference Gateway and Cloud Load Balancing alongside Mannequin Armor, NVIDIA NeMo Guardrails, and Palo Alto Networks AI Runtime Safety, all utilizing Service Extensions.
188. Cloud Load Balancing has optimizations for LLM inference, letting you leverage NVIDIA GPU capability throughout a number of cloud suppliers or on-prem infrastructure.
189. New Service Extensions plugins, powered by WebAssembly (Wasm), allow you to automate, prolong, and customise your functions with plugin examples in Rust, C++, and Go. Assist for Cloud Load Balancing is now typically obtainable, and Cloud CDN assist will observe later this yr.
190. Cloud CDN‘s quick cache invalidation delivers static and dynamic content material at international scale with improved efficiency, now in preview.
191. TLS 1.3 0-RTT in Cloud CDN boosts software efficiency for resumed connections, now in preview.
192. App Hub offers streamlined service discovery and administration by automating service discovery and cataloging.
193. App Hub service well being allows resilient international providers with network-driven cross-regional failover. Obtainable later this yr.
194. Later in 2025, you’ll be capable to use Non-public Service Join to publish a number of providers inside a single GKE cluster, making them natively accessible from non-peered GKE clusters, Cloud Run, or Service Mesh.
Then, that will help you safe your workloads, we launched enhancements to guard distributed functions and internet-facing providers towards community assaults:
195. The brand new DNS Armor detects DNS-based knowledge exfiltration assaults carried out utilizing DNS tunneling, area era algorithms (DGA) and different refined strategies. Obtainable in preview later this yr.
196. New hierarchical insurance policies for Cloud Armor allow you to implement granular safety of your community structure.
197. There are new community sorts and firewall tags for Cloud NGFW hierarchical firewall insurance policies, coming this quarter in preview.
198. Cloud NGFW provides new layer 7 area filtering, permitting firewall directors to watch and management outbound net visitors to solely allowed locations. Coming later in 2025.
199. Inline community DLP for Safe Internet Proxy and Utility Load Balancer offers real-time safety for delicate data-in-transit through integration with third-party (Symantec DLP) options utilizing Service Extensions. In preview this quarter.
200. Community Safety Integration, now typically obtainable, helps you keep constant insurance policies throughout hybrid and multi-cloud environments with out altering your routing insurance policies or community structure.
201. Imperva Utility Safety is built-in with Cloud Load Balancing, through Service Extensions, and is now obtainable within the Google Cloud Market.
Further accomplice bulletins
We’ve all the time taken an open strategy to AI, and the identical is true for agentic AI. With updates this week at Subsequent ‘25, we’re now infusing companions at each layer of our agentic AI stack to allow multi-agent ecosystems. Right here’s a better look:
202. Professional AI providers: Our ecosystem of providers companions — together with Accenture, BCG, Capgemini, Cognizant, Deloitte, HCLTech, Infosys, KPMG, McKinsey, PwC, TCS, and Wipro — have actively contributed to the A2A protocol and can assist its implementation.
203. AI Agent Market: We launched a brand new AI Agent Market — a devoted part inside Google Cloud Market that enables clients to browse, buy, and handle AI brokers constructed by companions together with Accenture, BigCommerce, Deloitte Elastic, UiPath, Typeface, and VMware, with extra launching quickly.
204. Energy brokers with all of your enterprise knowledge: We’re partnering with NetApp, Oracle, SAP, Salesforce, and ServiceNow to permit brokers to entry knowledge saved in these widespread platforms.
205. Higher area alignment and co-sell: We launched new processes to higher seize and share companions’ important contributions with our gross sales staff, together with elevated visibility into co-selling actions like workshops, assessments, and proofs-of-concept, in addition to partner-delivered providers.
206. Extra accomplice earnings: We’re evolving incentives to assist companions capitalize on the most important alternatives, resembling a 2x enhance in accomplice funding for AI alternatives over the previous yr. We additionally launched new AI-powered capabilities in Earnings Hub, our vacation spot for monitoring incentives and development.
207. We partnered with Adobe, the chief in creativity, to convey our superior Imagen 3 and Veo 2 fashions to functions like Adobe Categorical.
208. Along with Salesforce’s Agentforce, we’re main the digital labor revolution, driving large positive factors in human augmentation, productiveness, effectivity, and buyer success.
Safety
We provide important cyber protection capabilities for at present’s difficult menace surroundings, and launched a variety of new improvements:
209. Google Unified Safety: This resolution brings collectively our visibility, menace detection, AI powered safety operations, steady digital red-teaming, essentially the most trusted enterprise browser, and Mandiant experience — in a single converged safety resolution operating on a planet-scale knowledge material.
