The Cloud Goes Wild on AI as User Adoption Soars Past 70%

Artificial intelligence (AI) is taking over the cloud — with a study showing that 70% of companies using the cloud are now using managed AI services to some degree.

A new report released by Wiz Research analyzed 150,000 cloud accounts and found an explosion in the adoption of AI in the cloud, with 70% of organizations using managed AI services.

To put this in perspective, that makes AI almost as popular as managed Kubernetes services, used by 80% of the companies surveyed.

Microsoft was the indisputable winner of the analyzed providers, with Azure AI Services and Azure OpenAI leading in total AI deployment among significant cloud service providers — and AzureOpen AI usage increasing by 228% in 4 months.

The results highlight that AI adoption is shifting gears, with more and more enterprises experimenting with incorporating AI workloads into their cloud portfolios.

How AI is Revolutionizing Cloud Computing

More and more cloud providers have incorporated generative AI into their products, such as Microsoft’s Azure AI Studio, which was announced last November, and Google Cloud, adding generative AI support to its Vertex AI platform last June.

With AI hype in full swing, it’s becoming increasingly clear that cloud computing is a key enabling technology for developing an efficient insight economy.

Although many companies are investing heavily in AI, there remains a significant proportion who are still experimenting with the technology. For instance, 32% of companies have deployed less than ten instances.

Part of the reason for this “testing the waters” approach is that training and fine-tuning generative AI models can be extremely costly. The services themselves are not cheap to run, with some experts estimating that ChatGPT costs?as much as $700,000 per day to run.

For this reason, many organizations are limiting the number of instances deployed in cloud environments to see what value AI offers and whether it warrants further investment.

At this stage, the majority of interest centers around OpenAI – with 53% of cloud environments using OpenAI or Azure OpenAI SDKs, which integrate with services including GPT, DALL-E, and Whisper.

In their findings, Wiz said: “Generative AI is a truly cloud-native technology, with training and inference remaining highly compute-intensive for most use-cases, and therefore encouraging organizations to take full advantage of the computation and storage scaling that the cloud has to offer.

“Cloud service providers have stepped up to meet customer demand by offering both revamped versions of their existing AI and ML services as well as entirely new solutions for building with generative AI in the cloud.”

The Value of LLMs

So far in its development, generative AI has gained traction and interest based on its ability to improve productivity. Research from MIT shows that artificial intelligence can improve a worker’s performance by as much as 40% compared to workers who don’t use it.

At the same time, the versatility of language models like GPT-4 and Claude 2 opens the door to a high volume of use cases for employees to benefit from, from generating text and code to providing natural language summaries of isolated data signals to provide human users with greater context.

Considering that 60% of corporate data was stored in the cloud as of 2022, working toward the AI cloud allows users to extract insights from a wider range of data signals than ever before.

Despite these possibilities, many business leaders have ultimately remained uncertain about the value provided by the technology. For instance,

PWC reports that 88% of business leaders have listed achieving measurable value from AI as one of their top challenges. 85% have also listed the cost of adoption as a challenge.

So, for the foreseeable future, both end-user companies and cloud service providers will need to concentrate on improving the efficiency of training AI and machine learning models if adoption is to reach its full potential. This could be why so many companies gravitate toward managed services – to limit overall costs.

The Bottom Line

AI is here to stay. While generative AI solutions like GPT-4 are not perfect – they’ve done enough to capture the attention of the enterprise market. Whether it is economically feasible for the average company to run AI workloads in the cloud remains to be seen, but these early signs look promising.

There is clearly an active appetite to reap the productivity benefits of implementing AI, even if there is still a lack of clarity around the economics of doing so.