Reader poll – what next for AI and HPC

El Reg readers have their say on approach to building and deploying AI enabled applications and services

Sponsored Survey Last month we asked readers of The Register to answer a few questions about plans for artificial intelligence (AI) and high performance compute (HPC) in a bid to gauge the extent to which your organisations are engaged in building and deploying AI enabled applications and services and whether they have access to the secure, CPU intensive compute capacity needed to support those workloads.

You responded admirably (thank you) with 159 people sharing their thoughts. The results of the poll indicate that a majority of the companies are already actively involved in incorporating AI into their business operations in one way or another, but recognise there are barriers to overcome in either getting projects off the ground or scaling them up to the next stage of their evolution.

When asked which of the following best describes their organization's current approach to building and deploying AI enabled applications and services for example, almost half (49 percent) reported that they were either at the planning, deployment or scaling up stage of that journey, with an almost equal number (48 percent) currently not engaged in AI projects.

Question 1 Results

Source: The Register

The fact that so many of those taking part in our poll are yet to dip their toe in the waters may stem from lingering barriers to the adoption of AI and HPC technologies. We asked what were the biggest challenges facing their organization in implementing AI/HPC. Almost 43 percent of our readers cited a lack of AI skills and expertise, strongly indicating that they find it difficult to find staff with the requisite knowledge to drive momentum behind either new or existing AI/HPC projects.

Question 2 Results

Source: The Register

Just under 17 percent also indicated their organizations struggled to get management buy in for AI/HPC projects, which might suggest that senior executives might be concerned that complexity of the technology and its potential implementation costs may make it difficult to identify immediate value to the business. Sourcing access to sufficiently powerful compute/storage infrastructure to handle AI/HPC workloads was an issue cited by almost 15 percent.

Over a third (34 percent) of The Register readers reported difficulties in maintaining compliance with regulatory requirements and data protection legislation – a critical element considering the large volumes of data AI workloads typically process, often drawn from multiple sources. That finding was reinforced by responses to our third question, which asked which was the single most important factor under consideration when implementing datacentre/hosting infrastructure.

Question 3 Results

Source: The Register

Over 60 percent highlighted data security as the number one priority, which gives some insight as to the sensitive nature of the information being fed into AI workloads and stored in the HPC architectures responsible for running them. The breadth of response to this question dwarfed all other options. Scalability (seven percent), energy efficiency (seven percent) and hosting capacity (six percent) were identified as the critical consideration by a much smaller portion of the readers taking part in our poll.

Cybersecurity and data control consistently inhabit the list of priorities amongst IT decision makers, so it's no surprise that they extend into the relatively new realm of AI and HPC as well. Tech companies have already risen to this new challenge, developing new ways to protect data at rest, use and in transit to make sure that the systems they traverse are tamperproof. The Intel® Software Guard Extensions (Intel® SGX) capabilities embedded in the company's latest 5th Gen Intel® Xeon® CPUs for example are designed to enforce the isolation of sensitive information in a trusted execution environment at the application level to prevent access or modification from untrusted sources.

Intel has also worked hard at building new AI-optimized processors. The Intel® Advanced Matrix Extensions (Intel® AMX) accelerators in the new Xeon processors are designed to deliver GPU-level performance at CPU-level price points, while lower power consumption can be achieved throughout datacentre operations (including cooling and ventilation) to satisfy the requirements of sustainability conscientious AI software architects and application developers.

Sponsored by Intel.

More about

More about

More about


Send us news