Enterprise enthusiasm for cloud and its promise of streamlined, unified and agile operations has been tempered over the last year by failures to extract business value, according to a recent HFS Research report.
While nearly two-thirds of organizations have made strategic investments in cloud, fewer than one-third are realizing their ambitions, the report said.
The research firm collaborated with consulting firm EY to survey more than 500 senior executives from Global 2000 companies over a six-month period. It found widespread dissatisfaction, not with cloud itself, but with the technology’s business outcomes.
“Half of cloud-native transformations are abject failures,” Matt Barrington, emerging technologies leader at EY, said in the report, mostly when technology isn’t aligned with business objectives.
As cloud matures, the ROI clock is ticking. Massive modernization efforts across industries have succeeded in moving billions of dollars in enterprise workloads from on-prem data centers to more efficient hyperscaler infrastructure. While that push is expected to continue into next year, companies are anxious to see corresponding returns.
“Unfortunately, a lot of companies didn't realize that you need to spend a lot more time and attention figuring out how to optimize your data and processes, and make real changes to the way you operate,” Phil Fersht, CEO and chief analyst at HFS Research, said. “A lot of CFOs are turning to their CIOs and asking, ‘What was that all about?’”
From migration to transformation
Cloud transformation setbacks happen when an organization fails to meet its initial objectives, whether they are financial, operational or strategic, Ragu Rajaram, global cloud consulting leader at EY, said in an email.
The symptoms of failure are manifold, Rajaram said, identifying four pain points:
- Unexpected cost overruns
- Operational inefficiencies that lead to fragmented systems, impeding rather than streamlining operations
- New security vulnerabilities that increase cyber risk
- Skills gaps that prevent full utilization of cloud benefits
The culprit is often an unwillingness or inability to untether the enterprise from pre-migration habits.
“You have to go through some sort of transformation with your infrastructure, your data and your processes to do more than just layer on some new tech,” Fersht said.
As organizations overcome initial migration hurdles, the shift to cloud-first strategies is flattening the learning curve, the report found.
Two in 5 respondents said their organization has prioritized collaboration through cloud and more than one-third are leveraging cloud to speed product development and integrate digital business functions.
Still, more than one-third of respondents are in the process of moving workloads and establishing cross-functional workflows, and there's a small portion, 16%, who have yet to formulate a companywide cloud strategy.
The risk for these organizations is migrating technical debt to cloud, where it can drive cost overruns and other inefficiencies, Rajaram said.
“Large organizations are being held back by their technology and process debt,” said Rajaram. “Many create a parallel cloud-native organization next to brick-and-mortar heritage operations.”
The challenge is to weigh the cost of refactoring legacy applications against the relative ease of a lift-and-shift strategy.
“Refactoring optimizes applications, offering long-term benefits like performance efficiency and scalability, but at the cost of high initial investment, complexity and potential disruption,” Rajaram said.
Replatforming, the other migration option, is faster, cheaper and less disruptive. It also limits scalability and leaves technical debt stranded in cloud.
“There is rarely a one-size-fits-all or easy solution. Cloud-based technical debt requires a multifaceted approach that addresses underlying issues rather than treating symptoms,” Rajaram said.
AI raises the stakes
As CIOs grapple with technical debt, cost overruns, security concerns and skills gaps, generative AI has added a new variable to cloud strategies.
Most organizations will initially access large language model technologies through hyperscaler infrastructure, which is where most of their data resides.
“Generative AI is the first big tool that is very thirsty for infrastructure,” Fersht said. “If you want to scale out LLMs, which companies have been forced to do very quickly, you need a very strong cloud infrastructure to grow with you to support that technology.”
As generative AI gains an enterprise foothold, its appetite for data and infrastructure needs may promote sounder cloud strategy.
“This is the first time we’ve got a sophisticated software that business executives can use themselves,” Fersht said. “We’re seeing much more involvement from the C suite.”
Executives understand that generative AI can extract better insights faster, but there’s “a ceiling you’re going to get to if you don't design your processes and data to be optimized by the technology,” said Fersht.
Cost is another concern.
While bringing the models to the data is a far easier lift than moving huge stores of data, training and tuning models for enterprise-specific use cases won’t be cheap.
“Clients are now saying they need to invest in models, and they're going to expect cloud providers to deliver what they need without huge additional expense,” said Fersht. “I think you're going to have to develop real alignment between what enterprises want and what cloud providers can deliver.”
Fersht added, “Otherwise, this isn't going to end well.”
