• AI KATANA
  • Posts
  • AI discovers new rare-earth-free magnet at 200 times the speed of man

AI discovers new rare-earth-free magnet at 200 times the speed of man

Also: OpenAI to use Oracle’s chips for more AI compute

Author

AI KATANA
June 12, 2024

Good morning,

In today’s newsletter, we highlight groundbreaking advancements in AI and their far-reaching impacts. Materials Nexus has unveiled MagNex, a novel rare-earth-free magnet, developed 200 times faster using AI, offering a sustainable alternative for high-performance magnets. Gen AI is transforming workplace dynamics by automating mundane tasks, thereby enhancing job satisfaction and productivity. OpenAI's new collaboration with Oracle and Microsoft underscores the growing need for advanced compute resources in AI. Meanwhile, the surge in AI applications is driving up energy demands from data centers, posing significant environmental challenges. Nvidia's strategic software enhancements have substantially boosted the performance of its H100 GPUs, setting a new benchmark in AI training efficiency. We also explore top AI note-taking tools that are revolutionizing meeting efficiency, and AlphaSense’s remarkable $650 million funding round, elevating its valuation to $4 billion.

Stay informed and inspired!

Sliced:

  • 🪨 AI discovers new rare-earth-free magnet at 200 times the speed of man

  • 🤓 How Gen AI Can Make Work More Fulfilling

  • 🤝 OpenAI to use Oracle’s chips for more AI compute

  • ⚡️ How AI Is Fueling a Boom in Data Centers and Energy Demand

  • 🤫 Software tweaks are Nvidia’s secret AI sauce

UK-based deep-tech company Materials Nexus has leveraged its AI platform to rapidly develop a new rare-earth-free magnet named MagNex, achieving results 200 times faster than traditional methods. In response to the increasing demand for permanent magnets in electric vehicles and various technologies, and the environmental and geopolitical issues surrounding rare-earth mining, Materials Nexus's AI examined over 100 million potential material compositions. This innovation not only promises to meet the rising need for high-performance magnets without relying on rare earths but also significantly reduces costs and environmental impact. MagNex, synthesized with assistance from the Henry Royce Institute at the University of Sheffield, can be produced at 20% the cost and with 70% less carbon emissions compared to conventional rare-earth magnets. This breakthrough highlights the potential of AI in accelerating material discovery and could pave the way for sustainable solutions in industries reliant on magnetic technologies.

Gen AI has the potential to transform the workplace by not only enhancing productivity but also increasing job satisfaction. Research by Boston Consulting Group (BCG) shows that workers who spend less time on tedious tasks, or "toil," and more on activities they enjoy are less likely to leave their jobs. Implementing Gen AI tools can reduce toil by automating mundane tasks, thus freeing employees to focus on more fulfilling work. A study involving BCG’s administrative and HR professionals revealed that teams using Gen AI tools reported higher job satisfaction and effectiveness. Successful adoption of these tools hinges on managers who actively use and advocate for the technology, creating an environment where experimentation is encouraged. Co-creation, where employees are involved in tailoring AI tools to their specific needs, further boosts adoption and satisfaction. For instance, AI-driven calendar tools helped employees save time and enjoy their scheduling tasks more, while tools for professional development and focus work also led to increased joy and efficiency.

OpenAI is collaborating with Oracle and Microsoft to enhance its computing capacity for running AI applications like ChatGPT. This partnership allows OpenAI to utilize Oracle’s chip technology on Microsoft’s Azure AI platform, significantly expanding its infrastructure to meet growing demand. OpenAI's CEO, Sam Altman, highlighted the necessity for increased compute resources, indicating that Oracle’s chips will play a crucial role in the company's ability to scale its operations. Despite this new collaboration, OpenAI maintains that its core relationship with Microsoft remains unchanged, with the majority of its model training still occurring on Microsoft’s infrastructure. This development underscores the escalating need for advanced computing power in AI and the strategic alliances forming to address these demands.

The rapid growth of AI is significantly driving up the demand for data centers and, consequently, their energy consumption. Data centers, essential for storing and processing vast amounts of data, are witnessing exponential expansion due to the increasing deployment of AI technologies. This surge is not without substantial environmental costs; AI operations are extraordinarily power-intensive, with activities like training AI models consuming much more energy than conventional data processing tasks. As AI systems proliferate, energy demands from data centers are projected to double, equating to the consumption levels of entire countries like Sweden or Germany by 2026. This massive energy draw threatens climate goals, with tech giants like Microsoft already experiencing increased greenhouse gas emissions due to their AI initiatives. Efforts to mitigate these impacts include enhancing the efficiency of data center operations and pushing for more renewable energy sources.

🤫 Software tweaks are Nvidia’s secret AI sauce

Nvidia has significantly enhanced the performance of its H100 GPUs through a series of strategic software optimizations. These tweaks have enabled the H100 GPUs to achieve a remarkable 27% increase in performance in AI training benchmarks compared to last year. Key to these improvements are the use of the FP8 data format and Nvidia's proprietary transformer engine, which allows efficient, layer-by-layer precision management in AI models. This optimization reduces memory usage and speeds up data movement. Additionally, Nvidia has streamlined data transfers and balanced computational and communication tasks within GPU clusters to reduce overall execution time. These advancements have led to unprecedented scalability, demonstrated by Nvidia's ability to train a GPT-3 model with over 11,600 H100 GPUs at three times the performance of previous setups. Looking forward, Nvidia anticipates massive GPU deployments, with upcoming AI factories expected to utilize over 100,000 GPUs, driving the next wave of AI innovation and maintaining their competitive edge in the rapidly evolving AI landscape.

🛠️ AI tools updates

AI note-taking tools are revolutionizing how meetings are conducted and documented, promising to save time and improve efficiency. Executives, who often spend over 23 hours a week in meetings, can benefit greatly from these tools, which automatically transcribe discussions and highlight key points. Charter tested several AI note-taking solutions and identified top picks based on functionality and user experience. Circleback stands out for its superior summarization and speaker recognition in virtual meetings, though it is one of the more expensive options at $25 per month. Krisp, favored for its ease of use and affordability at $12 per month, excels in recording meetings directly from a computer's microphone and speakers without needing an in-meeting bot. For those who prefer a mix of manual and automated note-taking, Granola offers AI-assisted enhancements to user-generated notes, providing a blend of human input and AI efficiency. These tools not only streamline note-taking but also help distribute labor more equitably, reducing the burden of non-promotable tasks, which often disproportionately fall on women.

💵 Venture Capital updates

AlphaSense, an AI-driven market intelligence platform, has secured $650 million in funding, bringing its valuation to $4 billion. This funding round, led by Viking Global Investors and BDT & MSD Partners, also saw participation from prominent investors such as J.P. Morgan Growth Equity Partners, SoftBank Vision Fund 2, and existing backers including Alphabet's CapitalG and Goldman Sachs. AlphaSense, known for integrating AI and natural language processing to provide actionable business insights, aims to enhance its capabilities and expand its market reach with this new capital. The company also announced the acquisition of Tegus, a firm specializing in expert research and financial data, for $930 million. This acquisition will enrich AlphaSense’s platform with Tegus’s extensive private company data, allowing users to access more comprehensive insights. AlphaSense's recent milestones include surpassing $200 million in annual recurring revenue, doubling its revenue within two years, and expanding its global presence with a new hub in Singapore.

🫡 Meme of the day

⭐️ Generative AI image of the day