Delays in AI and ML projects due to GPU availability

New research from Civo highlights the significant challenges faced by AI and machine learning (ML) companies in accessing GPU resources, with 84.7% of organizations reporting delays in their AI/ML projects due to GPU availability, with over a third, (38.5%) having project delayed between 3-6 months.

  • 3 weeks ago Posted in

While supply chain pressures have alleviated since the pandemic, access to GPUs is not uniform, with a recent report from Bains suggesting this issue could worsen in the near future. AI companies are typically accessing GPU as cloud-based resources (31.6%), hybrid cloud resources (31.5%), or through on-prem infrastructure (13.5%)

The cost of running GPUs, either through provider pricing, inference, or power requirements, also presents barriers to capitalising on the hardware. In Civo’s research of nearly 600 AI professionals, 81.9% agreed that adequate GPU capacity is essential for maintaining long-term competitive advantage. Despite this, 96.9% of respondents cited high computational costs as a significant limitation on their organization’s ability to capitalize on AI and ML.

The research further found that a majority (74.5%) anticipate a growing need for GPU capacity over the next 12-24 months, yet many lack the resources to meet these expanding requirements. The top challenges companies cited in scaling AI/ML infrastructure were the high costs associated with infrastructure (54.3%), limited access to skilled personnel (43%), and the high cost of inference (40.3%).

Mark Boost, CEO of Civo, commented: “The transformative potential of AI is clear, but our research reveals that, for most organisations, the high costs and limited access to GPU resources are standing in the way of real innovation. For too long, the prohibitive costs associated with GPUs have been a bottleneck, forcing companies to put the brakes on projects that could otherwise drive industry-changing advancements. The benefits of Artificial Intelligence should not be solely reaped by companies large enough to both hoard GPUs and afford the energy costs of such scale.”

“At Civo, we believe there’s a better way forward. Our commitment to providing affordable GPU rates opens new doors for organizations, enabling them to scale AI initiatives without breaking their budgets. We’ve seen cloud innovation be hindered by restrictive pricing, and we should not let the same happen with AI.”

Distinctive advancements extend the Dynatrace platform to more audiences and use cases, unlocking...
ore42, G42’s digital infrastructure company, deploying AMD Instinct Accelerator technology in...
New integrations of NVIDIA BlueField-3 DPUs and NVIDIA Morpheus enable enhanced visibility and...
Driven by Insight technology experts, the report offers a guide for organisations looking to drive...
New research from Mendix, a Siemens business, has found that low-code platforms have become a...
Cisco is committed to helping service providers make their networks simpler, more resilient, and...
Study by Avature finds a mere 6% of respondents believe their technology stack fully supports...
New research from NTT DATA reveals that businesses are racing to adopt AI, yet a responsibility gap...