Michael Dell’s Post

This is when the #PowerOfFactory is leveraged at its finest to fuel the #AI and #GenAI innovation at scale. Any ideas on what will be deployed over that blue rectangle? 😉 #iwork4dell

Well done as expected.

The genAI revolution is here, and its all systems go for AI Factories. Dell Technologies’ commitment to supporting businesses in their genAI journey is clearly seen through our partnership with NVIDIA, resulting in an end-to-end genAI portfolio. With Dell AI Factory, our aim is to democratize the accessibility of genAI, bringing AI as close as possible to where data resides to minimize latency, lower costs, and maintain data security and governance within a controlled environment. The genAI revolution is here, and it’s up to organizations to choose if they want to disrupt themselves to stay ahead, or risk falling behind.

To build AI Factory for xAI

Let's build one of these to house the farmwave data coming from the "Farmbooks" in the farm fields around the world powered by Dell Technologies Ruggedized Tablets and built the smartest AI centric dataset for feeding the world. 😉 #Farmwave #AgTech #PrecisionAgriculture #AI

This is pretty awesome! Some interesting back of the napkin math, each one of those racks is going to contain 64 discrete top end GPUs and communication in rack alone will take about 25.6 TbE of networking. Each rack will require roughly about 100kW of power at full usage. With the rough amount of racks we see could train the GPT3 MLPerf benchmark in about 25 minutes. Amazing technology. #iWork4Dell #DellTechnologies

The change in the tech landscape anticipated by the adoption of GenAi is truly revolutionary. It is imperative for businesses to undergo transformation in order to survive, and this transformation will occur at a pace faster than anything we have previously encountered. #Iwork4delltechnologies

DSML trend: NVIDIA unveils new AI superchips and robots On March 18 Nvidia unveiled several major new AI products at their annual developer conference, including: 1.       The Blackwell series of AI chips – Blackwell is a platform for developing more powerful chips that can train massive AI models like GPT-4, yet it uses less energy than previous generations. This includes the B200 chip, the GB200 "superchip", and many others. 2.       Nvidia DGX Superpod - A shipping container-sized AI datacenter combining multiple racks of the new chips for enormous computing power, though extremely expensive. 3.       Project GR00T - A new AI foundation model specifically designed for controlling humanoid robots through natural language and observing human movements. 4.       Jetson Thor chip - Created to serve as the "brains" of robots working with GR00T to allow more autonomous capabilities. 5.       A quantum cloud computing service - This simulates quantum computers on Nvidia's chips to allow research and testing before accessing real quantum computers. The new chips and AI offerings aim to significantly increase the capabilities of large language models, image generation, and robotic control through more powerful and efficient hardware and tailored AI models. However, this enhanced performance comes with huge monetary and energy expenses, currently affordable only by major tech corporations. Apple, Oracle, Snowflake, SAP, Dell, and Nissan have all pledged to integrate NVIDIA platforms into their products and services. That’s my take on it: During the conference Jensen Huang mentioned that the rate of how computing power advances is insane. Notably, the NVLink Switch Chip boasts 50 billion transistors and can process 1.8 terabytes of data per second. Between the 1970s and 1990s, there were two AI winters, characterized by research stagnation and funding depletion. Yet, in recent years, AI research has experienced unprecedented leaps forward, with no signs of deceleration. As novel approaches conquer seemingly insurmountable challenges, progress becomes unstoppable.  Thomas Kuhn's concept of paradigm shifts is about scientific progress, but it can also be applied to engineering and technological progress as well. YouTube video that summarizes the presentation: https://lnkd.in/gT72fCgs

DSML trend: NVIDIA unveils new AI superchips and robots On March 18 Nvidia unveiled several major new AI products at their annual developer conference, including: 1.       The Blackwell series of AI chips – Blackwell is a platform for developing more powerful chips that can train massive AI models like GPT-4, yet it uses less energy than previous generations. This includes the B200 chip, the GB200 "superchip", and many others. 2.       Nvidia DGX Superpod - A shipping container-sized AI datacenter combining multiple racks of the new chips for enormous computing power, though extremely expensive. 3.       Project GR00T - A new AI foundation model specifically designed for controlling humanoid robots through natural language and observing human movements. 4.       Jetson Thor chip - Created to serve as the "brains" of robots working with GR00T to allow more autonomous capabilities. 5.       A quantum cloud computing service - This simulates quantum computers on Nvidia's chips to allow research and testing before accessing real quantum computers. The new chips and AI offerings aim to significantly increase the capabilities of large language models, image generation, and robotic control through more powerful and efficient hardware and tailored AI models. However, this enhanced performance comes with huge monetary and energy expenses, currently affordable only by major tech corporations. Apple, Oracle, Snowflake, SAP, Dell, and Nissan have all pledged to integrate NVIDIA platforms into their products and services. That’s my take on it: During the conference Jensen Huang mentioned that the rate of how computing power advances is insane. Notably, the NVLink Switch Chip boasts 50 billion transistors and can process 1.8 terabytes of data per second. Between the 1970s and 1990s, there were two AI winters, characterized by research stagnation and funding depletion. Yet, in recent years, AI research has experienced unprecedented leaps forward, with no signs of deceleration. As novel approaches conquer seemingly insurmountable challenges, progress becomes unstoppable.  Thomas Kuhn's concept of paradigm shifts is about scientific progress, but it can also be applied to engineering and technological progress as well. YouTube video that summarizes the presentation: https://lnkd.in/e97iXHNi

If you aren't thinking about your strategy around AI then you are at risk of being left in the dust by the competition. Discover how NVIDIA technologies are being used to build the AI factories that power the new era of accelerated computing and real-time #generativeAI. #GTC24