NVIDIA GTC 2024: We have the power, now we must tame it to build our AI-driven future

If there is one word to sum up this year’s NVIDIA GTC conference, it is “Power”. The raw, unadulterated power to build and run ever more powerful, capable GenAI models (and robots, digital twins, and virtual reality apps) that can transform our world in countless new ways. Between the new Blackwell GPU architecture and NIMs — portable LLM containers optimized for inference on NVIDIA hardware — we can now develop GenAI models and put them into production, at a scale hitherto unimaginable.

That is if you have the knowledge and capabilities necessary to turn these advances into impactful AI solutions. Therein lies the rub. NVIDIA GTC was a Mecca of hardware, technical innovations, and startups offering narrow features to facilitate individual steps in the AI lifecycle. It was not a conference of success stories and solutions. That lies in the future that we must create. We need to take the new power at our disposal and tame it. We need to develop new, more capable, specialized LLMs responsibly — to minimize business, legal, and, yes, societal risks. Above all, we need to orchestrate them as part of the complex pipelines of prompt templates, embedding models, vector stores, predictive AI models, upstream LLMs, downstream LLMs, guardrails, etc, that constitute real-world, transformative AI solutions.

This is where we see the future at Domino. NVIDIA has provided the power in spades. It has provided us with the engines to power the transformation of countless verticals and aspects of our daily lives. We just need to craft them into the reliable vehicles that will get us there. And we do that through responsible AI processes and platforms that integrate the ever-growing ecosystem of technologies that power transformative AI solutions.

Click here for more information about Domino at GTC, including information on how to unlock the power of NVIDIA infrastructure using Domino. You will also have access to the recorded GTC session on how to fine-tune LLMs using the NVIDIA NeMo toolkit and the Domino platform. (Free registration with Nvidia required.)