Cool stuff customers built, Sept. edition: Modern edge, faster blockchains, AI for nurses & more

AI and cloud technology are reshaping every corner of every industry around the world. Without our customers, there would be no Google Cloud, as they are the ones building the future on our platform. In this regular round-up, we dive into some of the exciting projects redefining businesses, shaping industries, and creating new categories. 

For our latest edition, we explore Uber’s revamped network with modern edge; smarter, AI-infused site search at Target; a wicked fast blockchain built by Keeta Network; HCA Healthcare empowers its nurses with a new AI-powered information portal; Deutche Telecom enhances its data science workloads; Google’s AI infrastructure delivers vastly improved inference for Baseten; Major League Baseball throws fans just the stats they want; and Kakao finds new ways to scale AI workloads.

Be sure to check back in October to see how more industry leaders and exciting startups are putting Google Cloud technologies to use. And if you haven’t already, please peruse our list of 601 real-world gen AI use cases from our customers.

Uber streamlines global network with hybrid edge

Who: Uber operates across six continents, connecting millions of riders and drivers while handling more than 100,000 concurrent trips and more than a million HTTP requests per second.

What they did: When Uber's existing edge architecture had sub-optimal routing paths, the company partnered with Google Cloud to redesign their global network approach. This work involved moving away from the distributed Envoy VMs and using Google Cloud's Hybrid Network Endpoint Groups (NEGs) instead. This new architecture directs traffic from Google's Global External Application Load Balancer — fronted by Google Cloud Armor for DDoS protection and Cloud CDN for caching — directly to Uber's on-premises infrastructure via Cloud Interconnect.

Why it matters: The results of migrating to Hybrid NEG-based load balancers were immediate. By removing all edge VMs, the traffic path became significantly more efficient, allowing Google's global network to handle the long-haul transit over optimized channels. This shift delivered a 2.6% latency improvement at the 50th percentile and 10% at the 99th percentile, directly improving service responsiveness. Removing the entire fleet of edge Envoy VMs resulted in significant cost savings, as well.

Learn from us: "At Uber, every millisecond defines the user experience for millions of people. By re-architecting our global edge with Google Cloud and Hybrid NEGs, we've created a more direct, lower-latency path for our services. This not only enhances today's user experience but also provides the high-performance foundation necessary for our next generation of AI applications, all while significantly reducing operational overhead for our engineering teams." – Harry Liu, Director of Networking, Uber

Target rebuilds search experience with AlloyDB AI hybrid search

Who: Target is one of America's largest retailers, serving millions of guests through its digital storefront and physical stores nationwide and has been a long-time pioneer of omnichannel retail experiences.

What they did: Target reimagined its site search using hybrid techniques that bring together traditional and semantic methods, backed by a powerful new foundation built with AlloyDB AI. Their new hybrid search platform combines classic keyword matching with semantic search to handle queries like "eco-friendly water bottles under $20" or "winter jackets for toddlers" that blend semantic nuance with structured constraints like price, category, brand, sizes or store availability.

Why it matters: Target has seen up to 10x faster execution, product discovery relevance improved by 20%, and half the number of 'no results' queries. Additionally, Target has reduced vector query response times by 60%, which resulted in a significant improvement in the guest experience while maintaining more than 99.99% uptime.

Learn from us: "Search at Target is evolving into something far more dynamic — an intelligent, multimodal layer that helps guests connect with what they need, when and how they need it. As our guests engage across devices, languages, and formats, we want their experience to feel seamless and smart." – Vishal Vaibhav, Principal Engineer, Target

Keeta Network hits 11 million transactions per second with Spanner

Who: Keeta Network is a layer‑1 blockchain that unifies transactions across different blockchains and payment systems, eliminating the need for costly intermediaries, reducing fees, and enabling near‑instant settlements.

What they did:  Keeta built a layer‑1 blockchain that unifies transactions across different blockchains and payment systems, eliminating the need for costly intermediaries, reducing fees, and enabling near‑instant settlements. The company engineered its network to meet the stringent regulatory and operational requirements of financial institutions using Spanner as the foundation for its distributed ledger, leveraging Spanner's availability and elastic scalability to scale up or down as needed without downtime or costly over-provisioning.

Why it matters: By facilitating cross-chain transactions and interoperability with existing payment systems, Keeta bridges the gap between cryptocurrencies and fiat, enabling a secure, efficient, and compliant global financial ecosystem. The company's architecture natively supports asset tokenization and digital identity, making it an ideal platform for stablecoins and real-world asset transfers while eliminating costly intermediaries and reducing fees for financial institutions.

Learn from us: "Keeta Network is unbounded by design, enabling it to scale horizontally to handle the increasing demand of participants. Similarly, Spanner’s scale-out architecture allows for linear read and write scaling in dozens of regions globally, all while maintaining consistency and latency." –Ty Schenk, CEO, Keeta Network

HCA Healthcare streamlines nurse shift changes with AI-powered handoff app

Who: HCA Healthcare is America's largest for-profit hospital network, operating 190 hospitals and approximately 2,400 ambulatory sites of care with 99,000 nurses across their system.

What they did: HCA Healthcare's Digital Transformation & Innovation department developed a tool for nursing staff, using Google's generative AI foundation models, that can ingest, analyze, and develop a cohesive and concise view of pertinent patient information for the oncoming nurse for handoff. The Nurse Handoff app shows nurses the electronic health record on one side and the AI-generated output on the other, allowing nurses to easily review and add information throughout their shift using hospital-provided mobile devices. 

Why it matters: This innovation addresses a massive operational challenge across HCA Healthcare's system: Nurse handoffs occur approximately 60,000 times daily across their facilities, with each handoff taking about 40 minutes per shift, totaling 10 million aggregate hours annually for all HCA Healthcare nurses. Nurses have rated the Nurse Handoff app as 86% factual and 90% helpful. The smoother the handoff, the more time nurses can spend with patients.

Learn from us: "One of DT&I's guiding principles for digital transformation is to address the administrative burden on clinicians. ...  We wanted to train a large language model system to organize information like a nurse and figure out key pieces of information that it needs to extract from the chart, then organize them in an intelligent way that the oncoming shift could easily consume. To transform patient care, the needs of caregivers must be prioritized, too." –Dr. Michael Schlosser, Senior Vice President and Chief Transformation Officer, HCA Healthcare

Deutsche Telekom transforms data science workloads with BigQuery DataFrames

Who: Deutsche Telekom is one of Europe's leading telecommunications companies, serving millions of customers across multiple markets and maintaining a significant presence in the global telecommunications industry.

What they did: Deutsche Telekom migrated their critical data science workloads from legacy on-premises PySpark systems to BigQuery DataFrames, executing a strategic two-phase migration that leveraged Gemini AI for accelerated code conversion. The team moved their CLV calculations and other distributed Python data processing workloads to BigQuery, standardizing on a single data processing technology while maintaining familiar pandas-like APIs.

Why it matters: The transformation eliminated performance bottlenecks from legacy systems, enabling faster time to insights, improved scalability, and reduced operational risk while allowing the team to leverage existing data science skills.

Learn from us: Deutsche Telekom completed their entire migration in just one person-week, with AI doing 95% of the heavy lifting on code conversion. Sometimes the biggest transformations require the smallest teams when you let technology do what it does best.

Baseten achieves 225% better cost-performance for AI inference

Who: Baseten is a six-year-old Series C company that partners with Google Cloud and NVIDIA to provide enterprise companies a scalable inference platform for their proprietary models as well as open models like Gemma, DeepSeek, and Llama, with an emphasis on performance and cost efficiency.

What they did: By leveraging the latest Google Cloud A4 virtual machines (VMs) based on NVIDIA Blackwell, and Google Cloud's Dynamic Workload Scheduler (DWS), Baseten has achieved 225% better cost-performance for high-throughput inference and 25% better cost-performance for latency-sensitive inference. Their approach is rooted in coupling the latest accelerated hardware with leading and open-source software to extract the most value possible from every chip, made possible with Google Cloud's AI Hypercomputer.

Why it matters: This breakthrough in performance and efficiency enables companies to move powerful agentic AI and reasoning models out of the lab and into production affordably. This unique combination enables end-users across industries to bring new applications to market, such as powering agentic workflows in financial services, generating real-time audio and video content in media, and accelerating document processing in healthcare — all happening at a scale and cost that was previously unattainable.

Learn from us: "Dynamic Workload Scheduler has saved us more than once when we encounter a failure. Our automated system moves affected workloads to other resources… and within minutes, everyone is up and running again. It is impressive — by the time we're paged … everything is back and healthy." – Colin McGrath, Head of Infrastructure, Baseten

MLB delivers real-time data to millions of fans with edge caching

Who: Major League Baseball is the world’s premier professional baseball league, overseeing 30 teams across North America and serving as the sport's governing body for everything from game rules to the massive data infrastructure that powers modern baseball analytics.

What they did: MLB switched from an outdated data system that frequently crashed to Memorystore for Valkey, which keeps all their live game data flowing smoothly to millions of fans. MLB can now deliver real-time stats, scores, and player information instantly to apps, stadium screens, and TV broadcasts.

Why it matters: MLB now handles nearly 10 billion requests daily and 15,000 API requests per second during peak games, ensuring that when a pitcher throws a 100-mph fastball or a runner steals second base, fans see the stats and updates instantly across every platform. In a sport where games can change on a single pitch, speed isn't just nice to have — it's essential for keeping millions of fans engaged and connected to the live action.

Learn from us: "With Memorystore for Valkey, we're in a position to move faster, build smarter, and deliver better experiences for everyone who depends on our data — from fans in the stands to broadcasters, analysts, and club staff." – Rahul Joshi, Principal Software Engineer, Major League Baseball

Kakao scales AI models with strategic JAX and Cloud TPU adoption

Who: Kakao operates KakaoTalk, South Korea's dominant messaging platform serving 49 million people – 93% of the country's population.

What they did: Facing critical limitations with their existing GPU-based infrastructure reaching power and budget capacity constraints, Kakao made a strategic shift to Google Cloud TPUs using the JAX framework. They worked with Google Cloud to adapt training tools for their specific needs, including customizing how they blend different data sources and process information to work seamlessly with their existing systems.

Why it matters: This enabled development of Kakao’s proprietary Kanana model family, with several Kanana models — including Kanana-MoE — recently released as open source on Hugging Face Hub. After switching to TPUs, Kakao observed an immediate and throughput increase of 2.7x across their models, along with improved cost-performance efficiency.

Learn from us: "We had two options: expand our GPU infrastructure and maintain our existing codebase, or adopt Cloud TPUs, which offered cost-performance advantages while requiring adoption of a new toolchain. We chose Cloud TPUs, viewing the short-term investment as worthwhile for long-term cost-performance benefits." – Minho Ryu & Nayeon Kim, Language Model Research Engineers, Kakao