What could possibly go wrong!? Well, probably nothing really (maybe?)
Leveraging “tens of thousands” of NVIDIA GPUs, Quantum-2 InfiniBand networking at 400 Gb/s, and NVIDIA’s nascent AI platform, the two firms are aiming to speed up the rollout and development of AI-based tools and apps powered by NVIDIA’s tech and Microsoft’s cloud footprint. This is the first time NVIDIA will have made the full stack of its cloud AI infrastructure available to the public, in partnership with Microsoft Azure.
With NVIDIA bringing its hardware clout to the table, Azure is bringing its unique blend of global scalability in virtual machine instances, which NVIDIA says will help speed up the training and deployment of all sorts of AI tools. NVIDIA says foundational models like Megatron Turing NLG 530B will see rapid advancement under the program, with the goal of developing “unsupervised” self-learning algorithms for building code, text, digital imagery, audio, and video. NVIDIA and Microsoft will also collaborate on Microsoft’s own DeepSpeed algorithmic refinement platform, which is designed to help AI instances self-learn more rapidly.
Scott Guthrie, EVP at Microsoft for Cloud + AI hailed the partnership, as the firms collaborate to explore the next wave in automated industry. “AI is fueling the next wave of automation across enterprises and industrial computing, enabling organizations to do more with less as they navigate economic uncertainties. Our collaboration with NVIDIA unlocks the world’s most scalable supercomputer platform, which delivers state-of-the-art AI capabilities for every enterprise on Microsoft Azure.”
Microsoft’s interest in developing AI platforms has been increasing exponentially in recent years. Microsoft has internal teams across the entirety of its portfolio exploring ways to incorporate AI self-learning algorithms across practically every product group. We’ve seen Copilot programming editing appear in GitHub and Visual Studio. We’ve heard how AI can enhance and speed up Xbox and PC video game development, handing off intensive tasks to AI. We’ve also seen piles of creepy AI-generated images that, while fun, also raise a question of copyright theft, given that AI train themselves using real-world art.
Indeed, as Microsoft’s investments in the space grow, there will doubtless be teething problems of every stripe. From copyright claims on content used to train algorithms, to morality questions over AI behavior. There are also hard questions about AI and automation putting the human labor force out of work as well. And hey, there’s always the chance the AI cloud computer becomes sentient and takes over the world, right? Well, maybe not.
However it all goes, it seems like Microsoft is poised to be on the cutting edge of this nascent tech, along with its enterprise partners on Azure.