Nvidia played a significant role in the surge of artificial intelligence last year, particularly with its high-end server graphics processors, such as the H100, which were instrumental in training and deploying generative AI models like OpenAI’s ChatGPT. In the domain of “local” AI tailored for personal computers or laptops in homes or offices, Nvidia is enhancing the capabilities of its consumer GPUs.
Recently, Nvidia unveiled three new graphics cards priced between \\(599 and \\\)999: the RTX 4060 Super, XX 470 Ti Super, and LTX 40080 Super. These cards come with enhanced “tensor cores” to support advanced AI applications. Furthermore, Nvidia is exploring the possibility of integrating these graphics cards into tablets produced by Acer, Dell, and Lenovo.
The upsurge in Nvidia’s overall sales and its market value surpassing $1 trillion can be attributed to the demand for the company’s GPUs, which are priced at tens of thousands of dollars each and are commonly part of an eight-GPU system.
Although Nvidia has traditionally focused on using GPUs for powering PC gaming, it has now tailored this year’s designs to handle AI workloads locally, eliminating the necessity to transmit data to the cloud.
The latest consumer-grade graphics chips from Nvidia are mainly intended for entertainment purposes but also have the capability to run AI applications. For instance, Nvidia asserts that the RTX 4080 Super can boost AI video production speed by 150% compared to its predecessor. Additionally, the new software upgrades introduced by Nvidia are anticipated to increase the processing speed of large language models by five times.
Justin Walker, Nvidia’s Senior Director of Product Management, emphasized the substantial base of powerful PCs for AI applications, with over 100 million Nvidia GPUs shipped.
Nvidia anticipates a surge in new AI applications harnessing the enhanced features of its GPUs in the coming year. Microsoft is set to launch Windows 12 later this year, which will effectively utilize AI chips.
The innovative chip can be employed for various tasks like background removal in video calls or image generation using Adobe Photoshop’s Firefly generator. Nvidia is also developing tools to assist game developers in integrating relational AI into their games to create dialogues for non-player characters.
The distinction between server and edge computing is becoming more evident with Nvidia’s latest graphics cards like the 4070 Ti Super. While Nvidia has been synonymous with robust server GPUs, it is now positioned to rival Intel, AMD, and Qualcomm in the realm of local AI. These companies have introduced new chips tailored for “AI PCs” with specialized components for machine learning.
The tech industry is exploring efficient methods for implementing relational AI, which requires significant computational power and could result in substantial costs when executed on cloud services. Nvidia’s strategic shift towards local AI signifies a transition towards more efficient and cost-effective AI processing.
Nvidia envisions applications that can leverage both local AI capabilities for quick tasks and cloud-based models for addressing complex challenges. The new graphics cards are crafted to adhere to import regulations, enabling their distribution to China and offering Chinese researchers and businesses an alternative to Nvidia’s most potent server GPUs.
