AI's Biggest Bottleneck is Cost: Companies Blame Stalled Projects on Overruns Three Times as Often as on Compute Shortages

Why AI Projects Fail: Cost Overruns Outpace Compute Shortages

Cost overruns are a major bottleneck for AI projects, with 38% of respondents to a recent survey citing cost as the primary reason for stalled or canceled projects. This is in contrast to compute shortages, which were cited by only 12% of respondents. Despite this, 79% of respondents expect their organization's AI budget to grow over the next year.

However, only 26% of respondents say their organization judges AI spending by a measurable financial return it could defend to its finance team. This suggests that many companies are struggling to manage the costs of their AI projects effectively.

One way to address this issue is to use open-source code and open interfaces, which can help companies change models and providers without rebuilding every integration. Commercial vendors can also support these systems in production, but the test is whether these choices improve the cost and quality of completed work.

Behind the Headlines: The Real Reason AI Projects Stall

Behind the Headlines: The Real Reason AI Projects Stall

The real reason AI projects stall is not due to compute shortages, but rather cost overruns. This is a major concern for companies, as it can lead to stalled or canceled projects. However, it's not just companies that are concerned - investors are also worried about the pace of future data-center construction.

One way to address this issue is to use open-source code and open interfaces, which can help companies change models and providers without rebuilding every integration. Commercial vendors can also support these systems in production, but the test is whether these choices improve the cost and quality of completed work.

Industry groups are trying to close the measurement gap by developing open standards for measuring AI costs and returns. This will help companies to better understand the costs of their AI projects and make more informed decisions.

What to Do About AI Project Cost Overruns

What to Do About AI Project Cost Overruns

Companies can take several steps to address AI project cost overruns:

• Evaluate model choice against real work, then measure cost alongside quality.
• Use open-source code and open interfaces to change models and providers without rebuilding every integration.
• Commercial vendors can support these systems in production, but the test is whether these choices improve the cost and quality of completed work.
• Industry groups are developing open standards for measuring AI costs and returns, which will help companies to better understand the costs of their AI projects and make more informed decisions.