The Road To AI Lies Through the Cloud: Google Cloud Next 2024 Highlights
Over 30,000 attendees and 2,500 partners flocked to Google Cloud Next 2024 in Las Vegas this week. Google Cloud Next 2024 also saw the company’s renewed resolve for generative AI optimization and AI in general. Here’s a recap of the most important highlights from the company’s flagship cloud event.
- Google Cloud wrapped up its annual Google Cloud Next 2024 on Thursday this week.
- Over 30,000 attendees and 2,500 partners saw Google Cloud Next 2024 register a 100% growth.
- Google Cloud Next 2024 also saw the company’s renewed resolve for generative AI and AI in general.
Google steered away from its usual San Francisco-based location for Google Cloud Next and conducted the annual cloud conference in Las Vegas this year. In addition to a new location, Google Cloud Next 2024 also saw the company’s renewed resolve for generative AI and artificial intelligence (AI) in general.
The search and advertising giant’s ability to cater to the exciting and emerging new technology stems from its early investment and research in AI and its presence in computing infrastructure through Google Cloud, whose loss-making days are behind it.
“One of the reasons Cloud is showing so much progress is our deep investments in AI. We’ve known for a while that AI will be the next technology to transform companies. Our investments in AI infrastructure and models help put us at the forefront of the AI platform shift. And we’re proud that today more than 60% of funded generative AI startups, and nearly 90% of gen AI unicorns are Google Cloud customers,” noted Sundar Pichai, CEO of Google.
Ram Palaniappan, CTO at TEKsystems, attended Google Cloud Next 2024 in Las Vegas and attests to Google Cloud’s success with startups. “The startup ecosystem leveraging GCP has grown tremendously,” he told Spiceworks News & Insights. “The demo grounds had more vendors and serious products. This shows that GCP has established itself as a leader in the new wave with AI.”
Consequently, Google’s cloud division and AI advancements are among the top three globally, and they can scale. However, from the multiple products and services unveiled at Google Cloud Next 2024, it seems the focus this year is efficiency.
Palaniappan also noted the “tremendous growth in the number of attendees, both customers and partners within Google Ecosystem.” More than 30,000 attendees and 2,500 partners saw Google Cloud Next 2024 register a 100% growth.
If you were among the rest who couldn’t attend, here’s a recap of the most important highlights from the company’s flagship cloud event.
Google Cloud Next 2024 Highlights
“Every product in Google’s catalog had an AI uplift,” Palaniappan noted. “Announcements were made for every product portfolio – modern Infra Cloud, Developer Cloud, Data Cloud, Security Cloud, Collaboration Cloud.”
1. AI infrastructure
Arm-based chip
Google’s infrastructure optimization efforts for AI now include Axion central processing units, its answer to AWS’ Graviton, Alibaba’s Yitian 710 server, and Microsoft’s Azure Maia 100 and Cobalt 100 chips.
Benchmarking data, architecture details, and other technical documentation for Google’s custom Arm-based silicon for cloud infra remain unwrapped for now, although we know that Axiom leverages Arm’s Neoverse 2.
Axiom chips are designed for general-purpose computing, and Google claims they are 60% more energy efficient than similar X86 chips while offering 60% better performance. The company also said Axiom offers 30% higher performance than Arm-based chips by AWS and Microsoft.
Palaniappan opined, “Google is catching up with AWS and Microsoft, which already have an Arm-based chip. This all leads to an evolving trend of cost optimization and margin squeeze.”
AI hypercomputer
Google also announced the general availability of TPU v5p, its most powerful tensor processing unit, which can scale to tens of thousands of chips. A single v5p pod contains 8,960 chips, almost twice as many as TPU v4. TPY v5p also delivers 2x as many floating point operations per second, 3x more high-bandwidth memory, and is 4x more scalable.
“AI Infrastructure gets a big boost with Google TPUs, NVIDIA GPUs (H100), NVIDIA GB200 support from 2025, AI-optimized storage (Hyperdisk ML), Dynamic workload management [Dynamic Workload Scheduler], etc., including Google Axion ARM chip for datacenter,” Palaniappan said.
See More: Mark Your Calendars: Top Tech Conferences in April 2024
2. Vertex AI updates
First up is the Vertex AI Agent Builder, a no-code service that enables clients to develop and deploy AI agents. It combines multiple large language models, developer tools, and Google Search.
The company has implemented retrieval augmented generation (RAG) and vector search for grounded results (i.e., displaying results sourced from reliable sources) and reduced hallucinations. Vertex AI Agent Builder leverages Gemini’s large language models. Google demoed multiple use cases, including retail and contract management.
Additionally, Vertex AI now features more AI models, including Anthropic’s new Claude 3 family of AI models, Gemini 1.5 Pro, CodeGemma (Google’s code assistance model), Imagen 2.0 (text-to-image tech), and more.
Notably, Gemini 1.5 Pro is now in public preview. “Its context window expands almost 1 million tokens, is multimodal, and is multilingual. You can process 1 hour of video and generate subtitles in multiple languages. This is huge and competes strongly against OpenAI Sora,” Palaniappan said.
Google has also expanded MLops capabilities in Vertex AI, now featuring Prompt Management for prompt-related operations, Rapid Evaluation for performance evaluation, and more.
Further, Vertex AI is integrated with BigQuery and Alloy DB to take on Databricks and Snowflake.
3. AI in Workspace
Google Workspace, the productivity suite billions of people rely on for day-to-day tasks, is getting more AI features. Some of them include:
- Voice prompts for email writing in Gmail, buffing up email drafts
- Templates for Spreadsheets, change notifications
- Enhanced organization capabilities in Docs
More importantly, Workspace is getting a new AI-powered video creation service: Google Vids. Currently available for limited testing, Vids is designed to work inside Docs and Sheets for a collaborative experience to cater to organizational needs, such as creating explainer videos, product decks, etc.
Google’s strategy is to take on Microsoft’s Copilot push with these integrated offerings with Gemini across its apps stack.
4. Hyperdisk Storage Pools
One of the lesser-talked-about updates is Google’s Hyperdisk Storage Pool, a block storage service to attain cost and compute efficiency. It allows devs storage pooling to be used across multiple workloads, especially AI inferencing.
“Hyperdisk ML accelerates model load times up to 12X compared to common alternatives and offers cost efficiency through read-only, multi-attach, and thin provisioning,” Google said. “It enables up to 2,500 instances to access the same volume and delivers up to 1.2 TiB/s of aggregate throughput per volume — over 100X greater performance than Microsoft Azure Ultra SSD and Amazon EBS io2 BlockExpress.”
Hyperdisk Storage Pools enable clients to reduce the total cost of ownership and management overhead and utilize resources well.
“If we look at Google’s AI strategy – its vision and ability to execute on all layers, right from infrastructure to platform to model to application and agent, is tremendous, and they are certainly the leading end-to-end AI platform company,” Palaniappan concluded.
What was your highlight of Google Cloud Next 2024? Share with us on LinkedIn, X, or Facebook. We’d love to hear from you!
Image source: Shutterstock
MORE EVENTS
