Alphabet Inc. (GOOGL) Earnings Call Transcript & Summary

Thank you, Aparna. When it comes to climate change, people, governments and corporations are more motivated than ever before, but they need access to the right information to measure and improve upon their ESG commitments. Google Search interest for how to reduce my common footprint has risen by 460% over the past 10 years. Today, Google Cloud is making it easier for our customers to make better sustainability decisions every single day. I'm pleased to announce that Google Cloud Carbon Footprint is now generally available and free for every user in the cloud console so that you can see and optimize your carbon impact. Shopify uses detailed emissions data to drive their mission to be the lowest carbon commerce platform. Earlier this year, we announced the general availability of Earth Engine on Google Cloud for better climate risk prediction and resilience. Natural Resources Canada, Earth Engine on Google Cloud helps map the tree canopy and support models for permafrost, crop status and water resources. With this data, provinces can monitor their land status and identify how to rehabilitate and transform these sites. Google Maps recently expanded our eco-friendly routing features to 40 new countries in Europe. I'm pleased to share that eco-friendly routing is coming soon to Google Maps platform on Google Cloud for developers and the transportation industry. Ridesharing and delivery companies, for instance, can embed eco-friendly routes into their driver apps. To provide customer sustainability data sets and solutions, we have over 20 partners with a Google Cloud Ready Sustainability designation and a new partnership with Dun & Bradstreet and Climate Engine. Now as we wrap this keynote, I want to share my appreciation for all of the people who are building and transforming their business with Google Cloud. On behalf of our teams, we're excited to develop technology with you today, to help each of you create a more amazing and a better tomorrow. Thank you so very much, and enjoy the rest of Google Cloud Next. [Presentation]

Hey, everybody, so I'm standing backstage at the Google Cloud's HQ. We're literally seconds away from kicking off a keynote for developer. I am so excited to share our predictions. Are you ready? Let's do this.

Hello, everyone. Please welcome Google Developers' Vice President, Jeanine Banks.

Hello. Welcome to Next '22. I'm Jeanine Banks, and I lead developer products and community at Google. My team and I love empowering developers to build innovations for the future, which leads me to the theme of Next this year. Today, Meet tomorrow. To tell you a little more about how we think about tomorrow, my friends and I, Google Cloud, will share its top technology predictions for where we believe the cloud is headed over the next 3 years. Each one of us will share the one prediction we believe will be true by the end of 2025. And we'd love to hear what predictions you all come up with, too. You can do that by responding to our original video or creating your own video on Shorts or any other social video platform. Just use the hashtag Google Cloud Predictions and tell us all about it. But before we get into all of that, I want to talk about our developer community for a minute. I get most excited about the incredible ideas and new products coming from our community as well as the opportunity we have to help developers learn, grow and build powerful systems and engaging experiences. Google's developer community is inclusive, one were cloud developers at every level of expertise are welcome while being challenged at the same time. And being part of Google's developer community creates true economic impact because Google Cloud started by professionals are some of the highest paid in the industry. This same cloud developer community fosters the creators. And by that, I'm talking about the students, the career switchers and anyone else hoping to become a developer, who's encouraged and supported to bring their creations to life. Just like we saw with our partner in Brazil, Soul Code Academy, and one of our newest professional data engineers Parceira. Take a look. [Presentation]

So you just love to see this story? This is why I'm excited to come to work every single day, and the potential of what we can build together with all of you. Speaking of exciting, we recently announced a new partnership with The Drone Racing League. We built new immersive learning experiences with DRL that will blow your mind. You can participate in the Google Cloud challenge, we can get hands-on with DRL's race data and Google Cloud Services. You can predict race outcomes, give performance tips to DRL pilots to help them smoke their competitors and learn, all at the same time. And you even get to compete for a chance to win a trip to the season finale of the League's World Championship. You can get started with the challenge right now. We look forward to seeing what you'll build next as we fly into the future of cloud together. Do you see what I did just there? Fly drones. Okay. Are you ready? Let's go. My prediction is by the end of 2025, developers will start with neuro-inclusive design. We'll see a 5x growth in user adoption in their first 2 years in production. According to the National Institutes of Health, up to 20% of the world's population is neuro-distinct, with the other 80% being neuro-typical. These 2 groups make up what is called neurodiversity. And neurodiversity describes the ways, people experience, interpret and process the world around them, whether in school, at work or through social relationships. And here at Google, we believe the world and our workplace needs all types of thinkers. What do you think about that, Jim?

That's right. One in 5 of us here are neuro-distinct. But what is neuro-inclusive design? Neuro-inclusive design is designed for cognitive and sensory accessibility. One of the things that makes participation in meetings accessible to me is the Raise Hand function. It allows me to pause the cadence of the meeting to provide an opportunity to share my thoughts. And we realized that this made meetings better for everyone. It creates more structure for everybody, which enabled visibility and broader range of ideas. So neuro-inclusive design starts with raising accessibility for neuro-distinct people like myself, but everyone benefits. Good design is already neuro-inclusive when implemented properly. As an example, when developing interactive and visual features, we need to consider how noise vibrations or pop-ups show up in design because it creates sensory simulation that leaves distraction. Design principles can be made neuro-inclusive when you plan thoughtfully for balance, proportion, unity, light, color, space and patterns. And here are some tips for developers: Number one, design with simplicity and clarity; number two, remove distractions or extra visualizations like pop-up windows; three, avoid really bright colors or too much of a single color; four, stick with the predictable and intuitive user flow; five, be thoughtful about the vibe you are setting. Do you need music or sounds to set the tone? Or is that just an extra element that can create distraction; and finally, six, stay away from pressure points requiring a quick reaction from the users. This can add unnecessary pressure. Back to you, Jeanine.

Thanks, Jim. This upfront design thoughtfulness is just so important. How often have any of you been pressured to ship something, but you knew you could have improved the user experience with just a little more time. At Google, we feel that ideation, user research, testing and marketing, disproportionately help teams launch more inclusive products faster. While these are standard practices in software development, where we made a conscious decision to have all types of thinkers included across these phases, our eyes open in so many new ways to be more inclusive. For example, closed captioning in Google Meet helps all of us process information better visually. It also helps people participating in meetings, where a different language is used. When you build simple, clear experiences with fewer distractions, your products will inherently drive greater user adoption. In fact, what did I say, I predict 5x more in your first 2 years in production. This is because developers like us will have built the longing directly into our products. So that's my prediction. Now I'm going to pass the mic to some of my friends here at Google Cloud to share their predictions. Thanks.

My name is Eric Brewer, and my prediction is that by the end of 2025, 4 out of 5 enterprise developers will use some form of curated open source. Now you're probably wondering what is curated open source. Curated open source is just open source as you know it but with a layer of accountability. What I mean by that is curated open source, comes with support for developers. And the curated in this case, will focus on not just finding vulnerabilities, but helping to fix them, too. They'll update all dependencies and track new ones. With curated open source, the curators who built in automation for testing may even offer response-based SLAs. And here's why this is so important for the community. Open source is everywhere. It helps power our electrical grids, water supplies, oil pipelines. It's fundamental to all clouds, most nations and even widely used in proprietary software. Open source is public infrastructure and a substantial part of our everyday lives. So now what? Open source is here to stay, but everything it powers is vulnerable. The incidents are real and they are costly. That's why governments are stepping in with regulations like FedRAMP and executive orders to combat cyber security. Regulations like these are super important and show us just how deeply the security vulnerabilities impact our lives. But in fact, these regulations are exactly why we need curation. Curated open source enables you to depend on the open source beyond the as-is approach we use today. We believe this philosophy so much that we are already working on. To help you build curated apps faster, we're releasing software delivery Shield, is the fully managed security solution that protect your software supply chain from source to deployment. And as part of SDS, we have our initial curated open store example called Assured OSS. This service curates open source packages used by Google and makes it available to you, our cloud developers. Google scan, analyze and fuzz test more than 250 Java and Python packages for security vulnerabilities on your behalf, and we'll update them as needed. So don't believe me that developers use some form of curation, how about this? Let us show you what we're doing at Google to make this reality. Hi, Aja.

Hey, Eric. So let's actually see what the software delivery life cycle looks like, if you use software delivery shield to enforce responsible use of open source through policy. So we're going to start with Cloud Workstations, a complete development environment in the cloud. Cloud Workstations is highly customizable. And the version I'm showing today has all the tools and compilers and everything I need, including the brand-new Source Protect extension. Actually, right here, you can see that Source Protect has flagged a dependency with a known vulnerability. I can fix this right now, online workstation before anything has ever checked in. Now Cloud Workstation is going to automatically detect those changes and it's going to redeploy my app on my workstations as needed. This shortens my dev loop and makes me more productive. So when I push my changes because I'm happy with them, Cloud Build is going to run continuous integration against our code base. Here, you can see a Cloud Build build report. And you can see that we provide SLSA Level 3-compliant build provenance. Cloud Build also scans for vulnerabilities. And here, we can see a list of vulnerabilities, including details on many of them. And in some cases, even the fix that we need to make to address it. Now it happens to be that the image that Eric and I are working with today has several external open-source dependencies. For example, Spring Boot Starter. In general, pulling these directly from the Internet could pose some risk. Fortunately, that dependency and many others are in the Assured OSS's portfolio so that I can use a version of that dependency that's been vetted by Google, which is what we've been telling us about.

This is exactly the point. You don't have to worry about these dependencies because they are already vetted for you.

And so now that I know that I don't have to worry as much about these dependencies, it's time to push the code to prod. So let's actually look at the delivery pipeline. Here's a cloud deployed delivery pipeline. Super simple one. We've just got a dev and a test environment. And when I'm happy with my code and dev, I can promote it to prod, and we can deploy the GKE. So while that's deploying, let's look at one last thing. Here's the GKE security posture page. Here, I can see security concerns on both the cluster and workload level. I can dig into any concerns if I need to, including seeing recommended actions to take to mitigate any issues that were identified. So that's the dev to prod tour of software delivery shield. So you can see how it helps you use open source responsibly, all the way from the -- when you start writing code to when it's released into production.

That was great. Thank you, Aja. We love co-creating. We just want to make sure there's an added layer of accountability to better support you and the apps you build. This is why we believe that 4 out of 5 enterprise developers will use some form of curated open source. Thank you. [Presentation]

Okay. First off, it is so nice to see everyone here in the room, given out for being back in person, right? All right. My name is Iman Ghanizada, and my prediction is that by the end of 2025, 90% of Security Operations workflows will be automated and managed as code. Okay, so SecOps teams are struggling to keep up with the attackers. We all know this, right? There's too much data, complex technology environments, and there's more adversaries now more than ever. I mean every day on the news, we hear about a new 18-year old has breached the company, right? So let's go ahead and let's pair this with the detection and response workflow, which is notoriously centered around toil. And it requires a linear growth in people to keep up with the volume of threats. But we all know we can just hire a bunch of more people, right? Like every CSO has $1 billion in their bank, and they're just waiting to hire 700,000 people. So this inefficient workflow has basically created the cybersecurity talent shortage. There's over 700,000 unfilled cyber jobs. And these jobs, they're high-stress, they're overloaded with toil-based work. And a lot of folks are, quite frankly, just burned out. So there's just no way that this issue is going to get solved if we keep doing things the way that we're doing them today. So the scale security across cloud, we're going to make security more agile and accessible to everyone through code. Here's how. So our Autonomic Security Operations Framework is designed to help you take advantage of our API first approach with Chronicle Security Operations and other tools, including many of those tools that are within our new Mandiant portfolio. The shift in our new framework takes traditional assembly line security operations workflows into a codified continuous feedback model we call Continuous Detection, Continuous Response or CDCR for short. You can imagine, it's kind of like CICD but for threat management. We've seen customers like BBVA and NCR and as well as our MSSP partners, like Cyderes, use our tools to build continuous detection and response workflows that scale across billions of alerts. And so earlier this year, we actually partnered with MITRE Engenuity, with Cyderes and others to launch Community Security Analytics. CSA is an open source repo that we created to the foster community collaboration on developing security analytics for your cloud workloads. These analytics can be deployed as code to complement the native detection capabilities in Chronicle and other Google Cloud tools. So it's kind of like having a team of devs collaborating on your detections. So I'll show you how to play these rules. But first, let's actually dive into an example, okay? Let's just say that there's a user outside of your DevOps team that gets access to impersonate a highly privileged production service account. That probably wouldn't be good, right? So yes. So Rule 2.20 analyzes your admin activity logs for permission grants on service accounts. And for this next part, we went ahead and prerecorded this because we want to save time and also I don't want to have you watch me fumble over my keys and take down USC or something. So first, we're starting in the terminal, and we've already gone to private repo here. Now we're going to go ahead and open up that URL rule. And we can add parameters to fine-tune it to our needs. So in case you haven't noticed, Chronicle's URL syntax is very, very, very simple compared to other detection languages. Now we're going to go ahead and commit and push our changes, and we already have a GitHub action set up to auto deploy these rules into our Chronicle instance. And by the way, Chronicle can be deployed as code and it can scale across petabytes of data without needing infrastructure or human involvement. I mean it's as cog-native as they come. Okay. So we're going to pivot into a Chronicle instance. We're going to refresh. And let's see, voila, okay, boom. So there's a new rule. And from this point on, it's going to start alerting when this malicious activity is identified. We can also run a retro hunt, which essentially takes this rule and it runs it against all of our historical data in Chronicle to see if there's an alert that we missed. And we've done retro hunts with customers that have taken seconds or minutes that may have taken them hours or days with their existing tools. I mean it's like it's uber fast. We can also use Chronicle SOAR to create an automation playbook, basically to figure out how to respond to this type of alert. And we also have APIs to automate this entire thing from ingest to analytics to response. So the moral of the story is what? None of us want to end up on the news, right? So in order to make this 90% prediction or reality, security analysts are going to have to work a lot more like devs, so they can free up time to focus on the most important threats to their organizations. So you all need to implement modern, developer-friendly workflows like CD/CR across your detection and response practice. And what I've shown you today is how we're making it possible for you to do so. Thank you.

My name is Kamelia Aryafar, and my prediction is by the end of 2025, AI is going to be the primary driver for moving to a 4-day work week. So what does this mean to you and me? A 3-day weekend. What it actually means is being able to comfortably complete 5 days' worth of work in 4 days or even less with efficiencies gained through AI. Enterprise use of AI alone exploded over the past few years, touching all aspects of business. One of the greatest reasons behind this is AI's huge potential to increase employee productivity. You have told us how excited you are to work with Google Cloud because we make all of the AI research, AI models and ML tool kits from Google available to you as enterprise-grade products and solutions, like Vertex AI. Since its launch, Vertex AI has helped data scientists shape their models faster into production by automating routine tasks like model management, monitoring and versioning. With Vertex AI, data scientists can now build and train ML models 5x faster, meaning increased time for experimentation, reduced custom coding and the ability to move more ML models into production. Today, with the announcement of Vertex AI Vision, we are taking this a step further and providing you with a fully managed development environment for creating computer vision applications. In the general keynote with a Smart City use case, my colleague, June Yang discussed how you can use Vertex AI Vision to reduce the time required to build and deploy competition applications from weeks to hours. Now let's dive into 3 areas that I, as a developer, I'm most excited about when it comes to Vertex AI Vision. The ability to use your own custom models, integration with BigQuery and developing external applications with SDKs. Let's see how. First in the Smart City example, we used a prebuilt occupancy analytics model to detect and count vehicles. Now if I want to do the same thing for bicycles, I can use a custom bicycle detection model in Vertex AI and easily import it into my computer Vision application. Basically, if the model works in Vertex AI, then it will also work in Vertex AI Vision. Second, I want to use the power of BigQuery to combine video annotations with other information in my data warehouse. By the way, I can also store annotations in the included Vision warehouse feature to easily search for insights across my video. By using Vertex AI Vision together with BigQuery, I can correlate traffic patterns with weather patterns or even make a forecast with BigQuery ML to predict future traffic patterns. Finally, I can use the SDK to access process video data and annotations and hook into a live stream of vehicle counts to power other applications or dashboards. It's that simple. And it can be applied to one video stream or even hundreds of video streams. This level of flexibility and scalability is unique to Google Cloud. And that's how we help you reduce the development time for computer vision applications from weeks to hours. Not just with Vertex AI Vision, but all of Google Cloud's AI products are built to help you be more productive and delight your customers. For example, using contact center AI, call center teams can manage up to 28% more conversations concurrently. That means a lot more productivity with Translation Hub, localization teams can translate documents into 135 languages in a matter of seconds, which means time saved for other efforts and a more inclusive workplace. Similarly, with Google Cloud's Recommendations AI, merchandising and e-commerce teams can now drive 40% more customer conversions. That means a lot more happy customers and a happy sales team, too. When you put together all of these productivity gains, powered by AI across the organization, a 4-day work week is a very distinct possibility. Thank you so much. [Presentation]

I'm Irina Farooq, and my prediction is that by the end of 2025, 90% of data will be actionable in real time using ML. I'm sure many of you are pretty skeptical of this prediction, and that's understandable. A recent survey found that only 1/3 of all companies are able to realize tangible value from their data. And we continue to hear that many of you are trying to fix that by taking on the operational burden of managing data infrastructure, moving data around and duplicating data to make it available to the right users and the right tools. So then how do we begin to overcome these barriers before data can be actionable in real time using ML? Since our inception, Google has been focused on delivering highly personalized information that is highly trusted by billions of people around the world. Data is in our DNA. The same data infrastructure that's allowed us to innovate is available to you, and that is why we believe we can make this prediction a reality. Take, for example, our customer Vodafone. As one of the world's largest telecommunication companies, Vodafone unified all their data so that thousands of their employees can innovate across 700 different use cases and 5,000 real-time data feeds. They now run AI development, 80% faster, more cost effective, all without compromising governance and reliability. So then how can we help you achieve your own data infrastructure vision? The short answer is in three parts. First, you can't act on data unless you can see it and trust it. Today, we're announcing automatic cataloging of all your GCP data with business context, in Dataplex, and you can integrate third-party sources to. What this means is that you no longer need to spend these looking for the right data and instead can spend time working with it. But once you find your data, how do you know you can trust it? Have you ever been in a meeting where someone questions the validity of a single data point and then nobody can trust anything that's being presented from that point onward? That's why I'm particularly excited about the new data quality and lineage capabilities in Dataplex, bringing intelligence and automation to help you trust your data. Second, you can't act on data unless you can work with it. To innovate, you've got to be able to use the best tools for the job across all your data. Speaking of the best tools, I'm excited about BigQuery's new support for unstructured data. Now you can be sure that you can -- your BigQuery skills will pay off across all your data from structured, to semi structured, to unstructured. It is also important to be able to use the best of open source tools. Last year, we introduced our Serverless Spark offering. And today, we're announcing that you can run Spark directly from BigQuery with fully integrated experience and billing. But this is still just the beginning. We have a bold vision for our Spark offering to bring the Google infrastructure magic, all without forking open source. Take, for example, mindmouth. The shuffle service powering [indiscernible] and data flow that helps deliver scale, reliability and performance that you know and love with those services. That's coming to a Spark's job soon. Lastly, you can't act on today's data tomorrow. We've heard from many of you that you struggle with making real time in context experiences, a reality for your own customers. Data flow are streaming analytics service, powers critical Google services, and we believe it can do the same for you. With data flow, you can use Apache bean to build the unified batch in real-time pipelines. You can start small, while having the assurance that you can process real type events at extreme scale if your application needs it. To summarize, when you can see the data, trust the data and work with data as it's collected, we can see how 90% of data will be actionable in real time using ML and the incredible innovation that, that will unleash. Thank you very much. [Presentation]

I'm [ Andi Gutmans ] and I predict that by the end of 2025, the barriers between transactional and analytical workloads will disappear. Traditionally, data architectures have separated these mixed workloads, and for good reason. Fundamentally, the underlying databases are built differently. Transactional databases are optimized for fast reads and writes, while analytical databases are optimized for aggregating large data sets. Because these systems are largely decoupled, many of you are struggling to piece together different solutions to build intelligent data-driven apps. For instance, to provide personalized recommendations for e-commerce apps need to support both transactional and analytical workloads on the same data set and without negatively impacting performance. At Google Cloud, we're uniquely positioned to solve this problem because of how we've architected our data platform. Our transactional and analytical databases are built on a highly scalable, disaggregated compute and storage system and Google's high-performance global network, allowing us to provide tightly integrated data services. And to help you unify your data across your apps, today I'm excited to tell you more about new capabilities we have recently announced. First, data stream for BigQuery, which allows you to easily replicate data from transactional databases into BigQuery in real time. Next, Database Migration Service, which provides a one-click migration from Postgres into AlloyDB for operational analytics. And lastly, we support Query Federation with Spanner, Cloud Sequel and BigTable write from the BigQuery console to analyze data in transactional databases. But don't take my word for it. Let's see how some of these technologies remove barriers for our fictitious company, Symbol Bank. Symbol wanted to integrate their core banking features in their app with market data to provide personalized real-time investment dashboards. The problem was their apps back-end was optimized for transactional workloads. So how do they maintain the responsiveness of their app? While adding analytical goodness, Symbol Bank chose Google Cloud's new fully managed Postgres compatible database, AlloyDB offering the capability to analyze transactional data in real time. Now migrating all their existing data with minimal downtime felt like a big lift for the Symbol engineers. But it turns out that database migration service makes this simple. And I'll walk you through just how easy it is. Database migration service that you migrate from Postgres to AlloyDB with continuous replication, minimizing downtime. Once we define where we're migrating from and where we're moving our data into, we can see the prerequisites for the migration directly in your UI. Sources are defined using profiles which contain host, user name and password, and you can predefine them like you've done here for Symbols Postgres incidents. Here, we define our destination and some basic configuration options and then we get to hit Create. This part will take a few minutes. So I've sped up time a little bit. Once it finishes, a quick test to ensure that it will all work and then hit Create and Start. Once the initial dump is finished, we're now in a state where we have both the old and new database populated with our live data continuously replicating from old to new. And that means we can do cool stuff like testing the performance of our new investment features against both the existing production Postgres database, and the new AlloyDB database side by side. Since AlloyDB is fully compatible with Postgres, you don't have to make any application changes. And as you can see, we're getting much better performance out of the new AlloyDB backend. AlloyDB is 4x faster for transactional workloads and up to 100x faster for analytical queries compared to standard Postgres, making it the perfect database for these kind of hybrid workloads. We all want to act on data in real time without the [indiscernible] of infrastructure, assembly and operations. We've given you a taste of how Google Cloud makes it easier for you to build data-driven apps on a unified platform. And this is why I predict that by the end of 2025, the barriers between transactional and analytical workloads will disappear. Thanks. [Presentation]

My name is Amin Vahdat and my prediction is, by the end of 2025, over half of cloud infrastructure decisions will be automated based on organizations usage patterns to meet performance and reliability needs. At Google, we believe the work we do today with our partners will define the next generation of infrastructure for the world. While some people look at infrastructure as a commodity, we see it as a source of inspiration. This inspiration comes from delivering capabilities not available anywhere else and pulling in the future by operating at a level of reliability and scale that might otherwise seem unimaginable. Our infrastructure is designed with the scale-out capability needed to support billions of users who use services like Search, YouTube, Gmail and our cloud services each and every day. We pioneered the model of entire buildings, operating as a single computing and storage system. Now with Spanner, we showed how services could run reliably at scale across the planet. And we've experienced network innovations by Google, Global Cash, [ D4 ] And Jupiter, shortening distances across the planet. This gave us the opportunity to reimagine what was possible from infrastructure in terms of scale and capability. Look at the world around us, the time for disruptive innovation has never been more profound. We're seeing incredible demand on the industry's infrastructure, yet simultaneous plateaus in efficiency. Year-on-year companies continue to push the boundaries of what infrastructure can provide, yet the burden of picking the just right combination of components continues to fall on you. To address this, we've engineered Golden Pass from silicon to the console. These pass combine purpose-built infrastructure prescriptive architectures and an ecosystem to deliver workload optimized, ultra reliable infrastructure. So let's talk about the investments we're making in infrastructure at Google in power and performance to make all of this possible. We partnered with Intel to codesign and build custom silicon [indiscernible]. This is called an infrastructure processing unit, or IPU, and it gives you massive performance and scalability to power high-performance data-intensive apps. These IPUs are at the heart of our new C3 VMs. C3s include the latest generation Intel Sapphire Rapids processor and custom designed offload based on the IPU that delivers 200 gigabit per second low-latency networking. And coupled with our new block storage HyperDisk, they can provide incredible storage performance. Now let me show you something else. From the small to the big meet the hardware behind the new Tensor processing unit, the TPU v4 platform. It's likely the world's fastest, largest and most efficient machine learning supercomputer. This liquid cooled board is a beast in both power and performance density. You see these pipes running across it, running chilled water over the board. They're there to ensure you extract the highest level of efficiency from the hardware, that allows secure isolated access and is at the cutting edge of services like natural language understanding, recommender systems and image processing. The TPU makes large-scale trading workloads up to 80% faster and up to 50% cheaper, compared to alternatives. When you talk about nearly doubling performance for half the cost, you unlock your imagination in terms of what just might be possible. The same IPUs and TPUs that power your services are the foundation that will enable us to automate over half of cloud infrastructure decisions in the next couple of years. They will support the telemetry data and ML-based analytics to proactively recommend the best infrastructure. It will be based on an understanding of how infrastructure balance points correspond to performance and reliability for your individual workloads. We don't think that you should have to think about hardware specifications. That is the last-generation cloud thinking. You will specify a workload and we'll quickly recommend, configure and place the best option for you based on your price, performance and scale needs. We know that these automated adaptive decisions deliver lower cost, more performance and higher reliability than any handcrafted solution. So it's my prediction that over half of cloud infrastructure decisions will be automated based on an organization's usage pattern, correct. Honestly, I think it's going to be much higher than that. It has to be in order to keep up with all the advancements you're making in technology. The burden and complexity of the infrastructure decision-making you have today will disappear through the power of AI and ML automation. And when you have freedom to focus on your solution delivery, the rate of innovation and customer benefits will only accelerate. While cloud has been transformative, we are still at the early stages. We're excited to continue to make the unimaginable possible and the possible easy. Thank you. [Presentation]

My name is [ Tarun Gemini ] and my prediction is that by the end of 2025, three out of four developers will lead with sustainability as their primary development principle. For the longest time, the focus was elsewhere. We needed to build it fast, build it securely, build it as simply as possible, build it at the lowest cost, build it reliably. Well, now it's also time to build it sustainably. We can't ignore the urgency required from all of us to meet climate targets. And while organizations are moving in the right direction, they struggle to take action. 65% of IT executives said they want to improve their sustainability efforts, but don't know how to do it. 36% of them said they didn't even have the measurement tools in place to track sustainability progress. So how can we help them out? We can give them better data about the environmental footprint of their business. So today, I am excited to announce that Google Cloud carbon footprint, which helps you measure, report and reduce your cloud carbon emissions is now generally available. Let's take a look, right from the cloud council, you can access the carbon footprint dashboard of your account. The underlying methodology is quite unique and is based on actual measurements of the energy used by machines in our data centers. It is also the complete picture of your emissions; not only emissions coming from electricity production, but also on-site fossil fuel emissions and other embodied emissions coming from data center hardware. This is also known as Scopes 1, 2 and 3. And of course, you can break down the data by Google Cloud Project, Google Cloud Region and Google Cloud Product. With a simple click, you can export your cloud emissions data to BigQuery for further analysis and to provide your sustainability teams with the data they need to report on your company's emissions. We also want to help you build new applications that emit less carbon by making the right choice at the right time. Let's say you want to deploy a new application to the U.S. West Coast. From a latency perspective, all of these options are very similar. Without more info, you might have picked Las Vegas, which happens to have a relatively carbon-intense electricity grid. But up in Oregon, you'd find a very clean grid full of hydropower and therefore, low carbon intensity. That is why it is indicated as a low-carbon region. In fact, a simple choice of Oregon over Las Vegas can reduce your gross electricity emissions of running that app by about 80%. And with this move, you not only saved carbon, you also saved money since the compute engine VM is cheaper in Oregon. A move that can help you save money and carbon is easy to make. When we tested this feature, we noticed new users were 50% more likely to choose a low-carbon region when they saw the icon, and that can make a very big difference. So how do we make it easy to identify more opportunities to lower your carbon emissions and deliver those options at scale? Well, active assist now shares recommendations to remove idle resources and their associated emissions. Actually, all of the sustainability features I just showed you today are embedded into Google Cloud Console and documentation. They are available out-of-the-box at no charge and for all developers. Sustainability is too important to be complicated. Before I go, here are two things to remember: One, moving to Google Cloud gives you the efficiency gains and energy benefits to reduce your emissions; two, when you build on the cloud, pick the region with the lowest carbon impact for your application. Because these tools are available, I believe that by 2025, three out of four developers will lead with sustainability as their primary development principle. Thank you. [Presentation]

Hey, everybody. I know I did jazz hands. Let's play with that weird energy, though, together here today. Good. So hi, my name is Richard Seroter. My prediction for you today is that by the end of 2025, over half of all organizations using public cloud will actually going to freely switch their primary cloud as a result of all these new multi-cloud capabilities available. So a quick story. I recently moved from the Seattle area to San Diego after a vacation this year with my family, we realized we needed something different. So bought a house in San Diego before I finally sold the house in Washington. And for a weird time there, I was multi-house and that was not good. But it wasn't -- it wasn't my desired end state, but that's okay. Sometimes you're in these multi situations while you transfer from one state to another. And that's related to my prediction here. I think in the years ahead, we're going to see companies use a multi-cloud strategy, not just as a way to hedge their bets but as a way to actually switch from their first cloud to their next cloud. So research data shows that the majority of companies are actually multi-cloud today, meaning they use more than one hyperscale cloud. It sounds probably pretty familiar to most of you. I'm personally talking to more and more companies, though, to using multi-cloud technologies as a way not to shift their -- just their workloads with their mind share to the different cloud as well. So let's see what that journey might look like. What would that look like look at like three different steps. We'll do some live demos, what can go wrong. So first, that first part of the journey is how do we meet you where you are today with what you got. You probably use another cloud. That's totally cool. Nobody is perfect. But here at Google Cloud, we've made some unique investments in a multi-cloud management plane that works with your compute and your data, even if it's on other clouds. So in this first step, you're starting to use Google Cloud, but you want to incorporate existing investments in other ones. So consistency matters a ton here and Anthos plays a huge part. Let's see how. So first off, you'll see here, I actually have an EKS cluster I built, and I've actually attached this to the Anthos control plane. From here, I can manage the cluster. I can view workloads, I can deploy things, I can troubleshoot. I can apply common policies. And even with our recent partnership and collaboration with the [indiscernible] Crossplane project, I can actually create and manage GKE clusters, AKS clusters, EKS clusters, provision and manage them all the same way from Google Cloud, which is pretty cool. Now what you do with your data, we talked about compute. If I think about my data, how do I analyze all this? We know that data transfer costs are real. Consolidation isn't always the right option. So for financial, strategic or even policy reasons, your organization might need data in different clouds. It's okay. So with BigQuery Omni, I can actually have my data lake in Amazon S3 or here in my Azure storage account. And I don't have to move the data to actually run my queries against it in BigQuery. I can actually analyze it where it resides without moving the data. So this is one of many cloud services that actually work wherever you are, whether they run across clouds or actually integrate with the different clouds. It's a big deal. So that's a starter phase, right? I'm building skills, I'm moving kind of comfort in this new home I have, but I haven't moved really anything. So some of you might think this is where you stop, right? Multi-cloud is just to use a bunch of clouds, but I don't think so. I think many are going a step further. So in the second stage, you start upgrading your tech where it is and growing your adoption of that secondary cloud. So step 2. As I start using Google Cloud services, let's say, things like GKE, which are awesome, I want to use it even more and more. So I might introduce things like Anthos clusters to Azure and AWS. This is the GKE API and GKE software running across clouds, which is amazing. And just now, we just shipped the capability where I can even upgrade those clusters in place on another cloud from the Google Cloud console. That's awesome. I don't have to jump all over the place to run all my infrastructure. Now here, you might also start creating a multi-cloud mesh. And say, look, I want to actually maintain services that run across clusters in different clouds. In this particular mesh, my web app is spread across GKE, EKS clusters, all of these different places I'm actually able to manage those services and see them and connect them wherever they are. Now let's talk about data. At this point in the game, you might use our great service data stream to redirect your existing Aurora cluster in AWS from Red Shift. Maybe you want to steer it a BigQuery instead. And I actually built a stream that connects from my Aurora database and feeds the data in real time to BigQuery. Pretty awesome stuff. They use data stream to move data around as well. So that's cool. So for some, multi-cloud is a bit of a phase. It's not a permanent state. You're not trying to get there. Your final stage could be a full-on migration to your new primary cloud. So once you've invested in Google Cloud, you might start taking advantage of really awesome stuff like GKE autopilot. This is a fully managed Kubernetes environment where we are running the whole thing. You don't have to manage clusters, you just deal with your workloads. It's awesome and exclusive to Google Cloud. Now what's really cool is you may also continue to use Anthos on Google Cloud to manage all of your fleets of GKE clusters at all of our regions around the world. I'm showing you some new dashboards here that are coming out. And this actually let me manage not just my GKE clusters but I'm managing on-prem, Bare Metal, VMware, Edge, doesn't matter. I'm getting a view of my entire fleet in managing that. So now you get things like BigQuery for analytics, you're in Google Cloud Spanner for distributed data, AI ML for amazing insights, all these unique developer tools and more. So look, if you forget everything else that I've talked about here today, remember that Google Cloud has a really unique management plane that meet you where you are, but honestly takes you further. So this is why I believe over half of companies using public cloud today will freely switch their primary provider in the coming years as a result of all these great multi-cloud capabilities. Thanks a lot. [Presentation]

Hi. I'm Yana [ Mundich ] and my prediction is that by the end of 2025, over half of all business applications will be built by users who do not identify as professional developers today. One of the most interesting opportunities as organizations evolve is that we will continue to see more development work in the enterprise taken up by teams and individuals outside of central IT. The adoption of no-code and low-code tools will unlock this potential by making the development process easier for more users. How many times have you been asked to do something but you had to say no? Because your road map or feature request list were already way too long. Well, with these tools, those business users, you had to say no to, can instead create apps and workflow automations themselves with no programming skills required. These no-code and low-code apps will be built collaboratively with developers like you who will provide the guardrails to keep the business secure while enabling business users to deliver their own solutions. And I'm not alone in thinking this prediction will come true. Leading tech analyst Gartner forecasts that by 2025, 70% of new applications developed by organizations will use low-code or no-code technologies. That's up from less than 25% in 2020. Organizations are getting ready for this change and our customers are already moving in this direction. Globe Telecom, a major telco out of the Philippines, reduced targeted business process turnaround time by 80% through experiences built by their citizen developers. And now it's demo time. Let me show you how Google Workspace's Appsheet is making all this possible today and in the future. Now I'm going to be Ann Gray, a business analyst, trying to help my team save time managing request approvals. Currently, the process is manual and disorganized, spread across ad hoc e-mails and chat messages. So to fix that, first, I'm going to build a no-code request approval app. And then I'll show you how my team and I can use this app to efficiently manage our development workflow, or approval workflow. With AppSheet, I have a single place to store my data and build my apps. I can also connect to other easy-to-use data sources like sheets or with the help of IT, I could connect to CloudDBs. Let me show you the solution that I could build as a business user. Here's my database. AppSheet helps me structure my data and prep it for app building. When I'm ready, I can create a new application with a single click. This creates a prototype app that will be usable on any desktop or mobile device. After some customization, here's what I've built. I have a new request for view for users to make requests and an approval view for approvers to do their thing. Each of these views is available to me in mobile apps and desktop apps by default, and I can also configure them to show up in Gmail and chat meeting my users where they are. Here is my Google Chat app. With this, my users can make request directly on their Teams chat space, and it's for using the same request for you I configured earlier. Here is my automation that will send the approval view to the approvers in e-mail. When I'm happy with my app, I can share it with my users, and I can add it to my chat spaces. Throughout this whole process, everything I created is controlled by governing policies set by the company's workspace admin. For example, here's what happens if I try to share the app outside of my domain, it actually blocks me and keeps company data secure. This allows IT to do two things. One, keep a clear line of sight on all apps so they can deprecate, update and retire them when necessary. And two, restrict access to only those users who need it. Now let me show you how my colleague Jeffrey and I can use this app to manage our request and approval workflow. I'll be Jeffrey now. I'm on Ann's team and I have a request to make. Let me ask Ann how to use her new app. She's [ added the button at the space ] and now I can quickly submit my reimbursement request. Okay. Now I'll be Ann again. I got this e-mail for Jefferies request. I can review and approve it directly from here. And if I have a pilot request to review, I don't have to click through individual e-mails. I can pop into the app and see everything in one place. This app is accessible to all the users, requesters and approvers. With simple sheets like expressions, it's configured to the requester to see only their own requests and approvers to see all of the information that they need. You saw just how easy we are making it for nontechnical users to create business applications that meet their immediate needs. With no code and low code, you and your business users now have more tools to work with. This is why I believe that over half of our business apps will be built by users who don't identify as professional developers today. Thank you.

Well, folks, that's a wrap on our predictions. We're excited about the conversations this will start and continue with all of you. Remember, if you want to share your own prediction, use the hashtag Google Cloud Predictions to tell us all about it. We can't wait to hear from you. Thank you for your continued inspiration and for partnering with us to build what's next. So are you ready? Enjoy the rest of this out. Thank you, everyone.