Profound technology shifts in history: Thomas Kurian, Google Cloud

Thomas Kurian, Google Cloud CEO, presented the opening keynote at the Google Next 2023, San Francisco, USA. We are now in the midst of one of the most profound technology shifts in history. Generative AI is a type of AI that can create new content, such as text, images, codes, feature audios, videos, etc. It can learn the patterns and structures of existing data, and generate new data with similar characteristics.

Google's generative AI is now delivering real results. It can automate, generate and transform! Many customers are announcing AI innovations at Next 2023. We are also announcing new generative AI products to help each one of you use those. We have world's leading customers and developers build AI with AI-optimized infrastructure. To help every person use AI, we have Duet AI. Third, to bring together the whole AI ecosystem, we can talk about our AI infrastructure, comprising Vertex AI and Duet AI.

We offer our AI cloud infrastructure in our cloud regions, in your data centers with Google distributed cloud infrastructure. This includes Google Compute Engine and Google Kubernetes Engine, leading AI software, frameworks and libraries, ultra-scale infrastructure, etc. Today, customers seeking sophisticated infrastructure use Google Cloud.

Example, Yahoo is migrating over 500 million mailboxes and 550PB data. In India, Mahindra & Mahindra relied on Google infrastructure to sell 100,000 SUVs that were sold in 30 minutes. The Government of Luxembourg is bringing digital services to their citizens using Google Distributed Cloud.

As generative AI evolves, so does the infrastructure requirements. We offer most advanced AI-optimized infrastructure. We offer AI-optimized storage, scalable networking, offloads, water cooling, etc. We make it easy for everyone to consume, with support for VMs and containers.

Today, we have made major advances. We are announcing Cloud TPU v5e. It offers 2x training performance/dollar, and 2x inference performance/dollar. We are offering custom-design hardware that supports complex AI models at a fraction of cost. Next, customers use Google AI engine for workloads. GKE (Google Kubernetes Engine for AI) Enterprise is meeting demands of today's most mission-critical apps. Customers are already seeing 45% productivity gains. It is reducing software deployment times by over 70%.

Third, we are announcing the general availability of our A3 VMs, powered by Nvidia H100. A3 offers three times faster training, and 10 times more network bandwidth than A2. Fourth, we are announcing the Titanium System, a tiered offload architecture. Titanium enables C3 VM block storage, to deliver 25% higher IOPS. Fifth, we are announcing general availability of next-generation Google Distributed Cloud for connectivity, sovereignty support, and regulatory requirement support. Sixth, we are introducing a brand new service -- Cross-Cloud Network, for security and speed across clouds. Over 70% of generative AI clients are Google Cloud customers.

We have also invested heavily across the AI infrastructure stack, and forged long-standing partnerships and collaborations. We are announcing an expanded partnership with Nvidia. We are supporting Nvidia GPUs that offer significant performance improvements.

Nvidia to the fore!

Jensen Huang, Founder, President, and CEO, Nvidia, said the two companies with perhaps, the deepest computer science teams in the world, are joining forces to reinvent the cloud infrastructure for generative AI. We are now starting from every single layer. This is the re-engineering of the entire stack. We are excited about the work that we are doing.

We are also going to put Nvidia DGX Cloud in GCP. DGX Cloud is powered by Base Command Platform, a comprehensive AI workflow management SaaS for AI developers that spans cloud and on-premises. It is an AI supercomputer in the cloud, designed for enterprise users with demanding needs and deep pockets. This will instantly benefit those working in the GCP. Third, we are looking at accelerating Vertex AI. The amount of work that we are doing in AI software is really incredible.

Generative AI and using large-language models is a breakthrough and cutting-edge. This is a whole new way of doing computing. We are working with Google to reinvent the software. LLMs can be distributed across giant infrastructures, and save time for AI researchers, scale up to gigantic next-generation modelsetc. All of that requires cutting-edge computer science.

Nvidia is now announcing PacsML. This is really a ground-breaking stuff. Second, our teams are starting to build the next-generation processors and AI infrastructure. Google and Nvidia are working on the next-generation AI supercomputer, called DGX GH200 based on a revolutionary new chip. It offers massive shared memory space of 144TB across 256 Nvidia Grace Hopper superchips.

Vertex AI and others

Kurian added that Fox Sports delivers moments that matter. Vertex AI has seen 150x rapid growth across gen AI projects. SaaS partners UKG is helping create new job descriptions. The Estée Lauder Companies Inc. is creating improved shopping experiences.

Our PaLM 2 upgrades 4x input token length. Cody delivers upto 25% improvements for code generation. GitLab uses PaLM 2 to explain vulnerability feature. Imagen significantly improves quality of generative images. We also offer Sec-PaLM for cyber security and response to any threats. We offer Med-PaLM for healthcare, where you can build models more accurately and safely.

Google Vertex AI is built for full data control with zero data leakage. We are also expanding Vertex AI with search and conversation. You want Google search for enterprise data. We are also announcing an enterprise Grounding Service for getting factual results. We are further announcing Embeddings.

Brazilian retailer, Magalu, is represented by Lu, a digital influencer with over 30 million social media followers. It uses search and conversational AI to develop Lu's Brain that can share recommendation tips and ideas to customers. Six Flags is helping 20 million visitors to their theme park each year, and have even more fun with a digital concierge. We are further enhancing Vertex AI with Extensions that connect your model with third-party apps.



Duet AI is also expanding capabilities. It will be generally available later this year. An example is Duet AI with Google Workspace. Workspace now has 10 million paying customers. He also talked about WorkDay. We appreciate being on this transformative journey with you, and building a new way to cloud.

Customers and partners

From startups to enterprises, new customers have announced how Google Cloud is accelerating their technology transformations through the use of generative AI. These include:

• AI21 Labs is partnering with Google Cloud to integrate industry-specific gen AI capabilities into BigQuery. The company is also leveraging Google Cloud’s AI infrastructure (TPUs and GPUs).
• Bayer is partnering with Google Cloud on generative AI, including using Vertex AI Search and exploring the use of Med-PaLM 2, a medically tuned LLM.
• Culture Amp, Australian unicorn and global employee experience platform, will use Vertex AI to help customers respond to employee feedback at scale and with speed.
• Fox Sports has entered an expanded partnership with Google Cloud to bring gen AI to major sporting event broadcasts using Google’s AI and search capabilities.
• GE Appliances is using Google Cloud’s gen AI to offer its users the ability to generate custom recipes based on the food in their kitchen and a conversational interface for appliance use and care.
• GM is exploring chatbot and document analysis gen AI use cases with Google Cloud. GM has successfully deployed Google Cloud conversational AI technologies in its OnStar service.
• HCA is announcing an AI collaboration with Google Cloud to explore Med-PaLM in support of doctors when treating patients.
• IHOP is using Google Cloud’s Recommendations AI technology to provide customers with better, more personalized recommendations on its online ordering platform.
• Meditech is using gen AI to empower employees and ultimately better serve its healthcare provider customers.
• MSCI is using Google Cloud’s gen AI to build solutions that will help investors identify and manage risk and opportunities, speed decision making, and build more climate-resilient portfolios.
• Runway has chosen Google Cloud to build, accelerate, and manage model deployment and bring gen AI to more creatives worldwide.
• Six Flags is using gen AI to produce an in-park infobot that will help guests navigate their visits from what rides have the shortest lines to what restaurants fit their needs.
• The Knot has rolled out Duet AI in Google Workspace as part of trusted tester program that gives employees early access to test and share feedback on Duet AI in Google Workspace.

Google has expanded partnerships with SAP and Docusign to apply gen AI to more business use cases:

• SAP is working with Google Cloud to build new industry solutions in automotive and sustainability, utilizing vast data customers have stored in SAP systems alongside Vertex AI.
• Docusign is piloting a new “smart contract assistant,” powered by Vertex AI, that will enable users to ask natural questions about complex contracts and documents.