Today, Pure Storage announced the release of AI-Ready Infrastructure (AIRI), a major move in serving the artificial intelligence (AI) and machine learning space. The newly announced solution combines Pure Storage FlashBlade and NVIDIA DGX-1 technology. According to Pure Storage, it is the industry’s first integrated AI-ready infrastructure for deploying deep learning at scale.
With AI and machine learning still relatively nascent technologies, the first question that comes to mind might be, “Why an integrated solution, and not just a reference architecture?”
I can give you three answers:
- For the companies with AI and machine leaning initiatives, those initiatives are often critical to the business.
- These solutions need lots of data and lots of speed.
- And existing infrastructure options seem to be holding organizations back.
According to ESG’s research on current machine learning and AI trends, the most commonly identified barrier experienced with AI and machine learning projects is the cost of the infrastructure required. In other words, the contemporary crop of solutions are not optimized for what AI and learning workloads need, so companies have been throwing money at the problem.
With Pure Storage and NVIDIA’s combined solution, organizations are provided turnkey infrastructure optimized for AI and machine learning workloads. While the expected result is more efficient and simpler infrastructure, there is the potential for a much bigger benefit. Companies are investing quite a bit in AI and machine learning personnel and expertise. The last thing these companies need is for the equipment to hold these projects back. These experts need to spend their time working with data sets and optimizing algorithms, the highly technically and highly valuable work, not cobbling together disparate hardware components to find the right level of storage and performance.
At the end of the day, Pure Storage is making a bet that companies want their data experts to focus on their core competency, the data and the algorithms, and to leave the hardware tuning to someone else. And from where I am standing, it looks like a smart bet.