AI Weekly Rundown (June 01 to June 07)
Major AI announcements from Nvidia, Intel, AMD, OpenAI, and more.
Hello Engineering Leaders and AI Enthusiasts!
Another eventful week in the AI realm. Lots of big news from huge enterprises.
In today’s edition:
📢 Nvidia CEO drops a series of AI announcements
🏢 AMD’s new chip architecture strategy for AI data centers
🔊 ElevenLabs' Text-to-Sound AI wows creators
💻 Intel’s new chips push for AI-ready data centers
📦 Amazon’s Project PI detects defective products before shipping
☁️ Microsoft’s Aurora AI could transform weather forecasting
🚀 Cisco and NVIDIA launched HyperFabric AI Clusters
🤔 Tesla's AI ambitions on hold? Musk diverts chips to X & xAI
🛡️ OpenAI insiders raise concerns over oversight and safety
🎧 Stability AI releases a text-to-audio model
🤖 xAI to build the ‘gigafactory of compute’
📊 New study reveals key findings on young people's use of Gen AI
🧠 OpenAI reverse engineers the workings of AI models
🎥 New Chinese video generation model rivals OpenAI’s Sora
🥈 Nvidia is now the second-most valuable company, overtaking Apple
Let’s go!
Nvidia CEO drops a series of AI announcements
Nvidia CEO Jensen Huang revealed the company's ambitious plans for annual AI accelerator upgrades, targeting a broader range of industries to expand its customer base.
It will release the Blackwell Ultra chip in 2025 and the next-generation Rubin platform in 2026.
It is also releasing a new server design, MGX, to help companies like HPE and Dell bring products to market faster.
They are promoting the use of digital twins in its Omniverse virtual world, showcasing a digital twin of Earth for sophisticated modeling tasks.
Introduces Project G-Assist, an RTX-powered AI assistant technology that provides context-aware help for PC games and apps.
G-Assist uses voice or text inputs and game window snapshots to provide personalized responses based on in-game context.
Developers can customize the AI models for specific games or apps, and they can run on the cloud or locally on GeForce RTX AI PCs and laptops.
Nvidia partnered with Studio Wildcard for a tech demo using ARK: Survival Ascended, showcasing how G-Assist can help with quests, items, lore, and challenging bosses. Check out full keynote speech:
AMD outlined new chip architecture strategy for AI data centers
AMD CEO Lisa Su introduced new AI processors at Computex, including the MI325X accelerator, set to be available in Q4 2024.
The CEO announced the MI325X accelerator, which will be released in Q4 2024, and outlined the company's plan to develop AI chips over the next two years.
Introduced the MI350 series, expected in 2025, which promises a 35x improvement in inference performance compared to the current MI300 series.
The company also teased the MI400 series, slated for 2026, based on the mysterious "Next" architecture.
With AMD and Nvidia moving to annual release cycles, the competition is heating up to meet the soaring demand for AI semiconductors.
ElevenLabs' Text to Sound AI wows creators
ElevenLabs introduces Text to Sound, an AI model that generates sound effects, instrumental tracks, soundscapes, and character voices from text prompts. The tool aims to help film, TV, video games, and social media creators produce high-quality audio content quickly and affordably.
They have partnered with Shutterstock to fine-tune the model using their diverse audio library of licensed tracks. Users can generate sound effects by logging in, describing the desired sound, and downloading the best results.
Note: This tool doesn't have a content filter and can generate any raw content through conditional prompting.
Intel’s new data center chips handle demanding AI workloads
Intel has announced next-generation Xeon 6 server processors to regain the data center market share it had been losing to AMD. They come in two varieties. The larger, more powerful version is designed to run the computations necessary to generate responses from complex AI models and other tasks requiring increased horsepower. Intel plans to help companies modernize their aging data center systems with Xeon 6 chips so they can generate new digital capabilities.
Intel also revealed that its Gaudi 3 AI accelerator chips would be priced much lower than its rivals' products.
Amazon’s Project PI detects defective products before shipping
Amazon has launched Project PI, which uses AI to scan products for defects before shipping them to customers. This AI system combines computer vision to visually inspect items with generative AI models that can understand things like text on packages.
As products go through a scanning tunnel, the AI checks for damage, incorrect colors/sizes, or expired dates. If it finds a problem, that item is isolated to evaluate the defect. Project PI already operates in several of Amazon's warehouses across North America. The system catches millions of defective products daily before they reach customers.
Microsoft’s Aurora AI could transform weather forecasting
Microsoft has developed a powerful new AI foundation model called Aurora that can make highly accurate weather predictions. It is trained on over a million diverse weather and climate data hours. This allows it to develop a comprehensive understanding of atmospheric dynamics and excel at forecasting various weather variables like temperature, wind speed, air pollution levels, and greenhouse gas concentrations.
What sets Aurora apart is its ability to capture intricate details at high spatial resolution (around 11km) while being much faster and more computationally efficient than traditional numerical weather prediction systems. Aurora's flexible architecture and training on heterogeneous datasets enable it to adapt to different forecasting tasks and resolutions.
Cisco has unveiled HyperFabric AI Clusters in collaboration with NVIDIA
Cisco and NVIDIA announced Cisco Nexus HyperFabric AI Clusters, an end-to-end infrastructure solution for scaling generative AI workloads in the data center. It combines Cisco's AI-native networking with NVIDIA's accelerated computing AI software and VAST's data storage platform.
It is designed to simplify the deployment and management of generative AI applications for enterprise customers, providing centralized control across the entire AI infrastructure stack.
The Nexus HyperFabric AI cluster will be available for early customer trials in Q4 2024, with general availability expected shortly after.
Tesla's AI ambitions on hold? Musk diverts chips to X & xAI
Elon Musk instructed Nvidia to prioritize shipments of AI chips to X and xAI over Tesla, diverting over $500 million worth of Nvidia's flagship H100 AI chips that were initially reserved for Tesla.
This decision could delay Tesla's plans to significantly increase its acquisition of H100 chips from 35,000 to 85,000 by the end of 2024, a crucial part of Musk's vision for transforming Tesla into "a leader in AI and robotics."
Consequently, this move could frustrate Tesla investors who are counting on Musk to deliver on his promises regarding autonomous driving and Tesla's AI capabilities.
OpenAI insiders raise concerns over oversight and safety
Open AI researchers are concerned about the lack of proper oversight, the influence of profit motives, and the suppression of whistleblowers working on advanced AI technologies. They warn of risks ranging "from the further entrenchment of existing inequalities to manipulation and misinformation, to the loss of control of autonomous AI systems potentially resulting in human extinction."
They want AI companies to agree to four principles: refraining from enforcing non-disparagement agreements, establishing anonymous channels to raise concerns, allowing employees to share risk-related information publicly while protecting trade secrets, and not retaliating against whistleblowers.
Enjoying the weekly updates?
Refer your pals to subscribe to our newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Stability AI’s sound generator creates drum beats and instrument riffs
Stability AI’s Stable Audio Open can generate up to 47-second audio samples based on text descriptions. The open AI model is trained on data from 486,000 samples of royalty-free music samples. The tool enables users to generate drum beats, instrument riffs, and ambient sounds.
However, the AI model has its limitations.
It is unable to produce full songs, melodies, or vocals.
Its terms of service prohibit users from using Stable Audio Open commercially.
Its training data is biased toward the English language and specific music styles.
xAI to build the gigafactory of compute in Memphis
The AI startup seeks to build the world’s largest supercomputer in a multi-billion dollar project. The company plans to use this supercomputer to develop AI products, including its chatbot, Grok. The facility will be powered by Nvidia's H100 GPUs. The project aims to be operational by fall 2025.
The project will reportedly use Nvidia AI chips originally intended for Tesla, raising concerns about conflicts of interest. Moreover, Musk hasn’t yet delivered Grok 2, an advanced AI model that he had promised in May.
New Study reveals insights on young peoples’ use of Gen AI
The study directly involved young readers and examined the use of generative AI by use, ethnicity, age, gender, and LGBTQ+ identity. Key findings include:
50% of the survey respondents (aged 14-22) have used generative AI. However, only 4% use it daily.
For 53% of respondents, the use case for generative AI was obtaining information, while for 51%, it was brainstorming.
Black young people are more likely to use generative AI compared to their white peers. Reasons include getting information, brainstorming ideas, and assistance with schoolwork.
Young people of Latin origin are more likely than white people to use generative AI for multiple purposes, including image generation and getting help with their jobs.
Out of respondents who have never used generative AI, 34% believed it would not be helpful.
Among people never having used generative AI, LGBTQ+ young people are more likely to use it in comparison to cisgender and straight respondents.
41% of respondents believe that generative AI will have a positive as well as negative impact on their lives in the next 10 years.
OpenAI reverse engineers the workings of AI models
In new research, OpenAI has shared improved methods for finding a large number of "features"—patterns of activity in AI models that are human interpretable. They developed new state-of-the-art methodologies that allow scaling sparse autoencoders to tens of millions of features on frontier AI models.
It demonstrated smooth and predictable scaling, with better returns to scale than prior techniques. And they could find 16 million features in GPT-4. The research also introduces several new metrics for evaluating feature quality.
OpenAI has shared the paper, code, and feature visualizations to foster further exploration.
New Chinese video generation model beats OpenAI’s Sora
Kuaishou, a Chinese tech company, has introduced Kling, an AI model for video generation. It can make videos up to two minutes long at 1080p resolution and 30 frames per second, vs. Sora’s one-minute videos.
Kuaishou claims Kling correctly simulates the physical properties of the real world, including complex motion sequences. Using a diffusion transformer, it can also combine concepts and create fictional scenes, such as a cat driving a car through a busy city.
The model is currently available as a public demo in China.
Nvidia is now the second-most valuable company, overtaking Apple
Nvidia rallied to record highs on Wednesday, with it’s stock market valuation hitting $3 trillion and overtaking Apple to become the world’s second most valuable company. This comes after Nvidia made a series of major announcements in the past week.
However, Nvidia’s stock has surged 147% so far in 2024, with demand for its top-of-the-line processors far outstripping supply as Big Tech races to build out their AI computing capabilities and dominate the emerging technology.
Microsoft remains the world’s most valuable company, with a market value of approximately $3.15 trillion.
That's all for now!
Subscribe to The AI Edge and gain exclusive access to content enjoyed by professionals from Moody’s, Vonage, Voya, WEHI, Cox, INSEAD, and other esteemed organizations.
Thanks for reading, and see you on Monday. 😊