The compounded benefits of cloud and machine learning has been saved
The compounded benefits of cloud and machine learning
State of Cloud: Trends & Predictions
A blog post by Diana Kearns-Manolatos, senior manager; and Jay Parekh, senior analyst, Center for Integrated Research at Deloitte.
The analytical capabilities of machine learning (ML) combined with the scalability of cloud computing provide a powerful duo to achieve truly transformational changes. Currently estimated to be between $2 and 5 billion, the cloud ML market has the potential to reach $13 billion by 2025.1 This projected exponential growth is a testament to the growing confidence among executives in cloud-based ML technology to further accelerate their AI programs.
Based on our analysis of Deloitte’s State of AI in the Enterprise, 3rd Edition survey, published July 2020, organizations that are already implementing a cloud ML approach are more likely to recognize the potential of AI technologies to transform their organization and industry. In our recent article, “Time, technology, talent: The three-pronged promise of cloud ML,” we uncovered three key findings:
- 49% of cloud ML users2 said they saw “highly” improved process efficiencies (versus 42% of total respondents).
- 45% saw “highly” improved decision-making (versus 39% of total respondents).
- 39% experienced “significant” competitive advantage (versus 26% of total respondents).
These cloud ML adopters are defined as respondents who stated their organization is using cloud-enabled AI along with cloud-enabled ML platforms or tools. While they represent slightly more than 15% of the 2,737 respondents in the State of AI survey, cloud ML adopters are further along the maturity curve compared with nonadopters and are able to extract benefits from these technologies, with responses showing additional gains in decision-making, process efficiencies, and competitive advantage.
Characteristics of cloud ML users
Our analysis of the State of AI survey data has found that cloud ML users believe that AI technology is essential for business success. Cloud ML adopters are almost twice as likely, compared with the rest of the respondents (defined as “others”), to say that AI technologies have a critical strategic importance today, as well as two years from now. This translates to a much more effective and active AI program. There are certain characteristics that cloud ML users portray that help in delivering value to the organization through AI/ML; some of these include:
- More on-the-ground action. Cloud ML adopters understandably have a better track record, compared with others, when it comes to the breadth and scale of their AI programs. There is a 1.7x difference in the number of such organizations that stated they have more than 20 prototypes (12% in cloud ML users versus 7% others) and more than 20 cases of full AI implementation and deployment (15% in cloud ML users versus 9% others).
- Scaling AI programs across the enterprise. Many AI programs are challenged with taking successful prototypes and pilot implementations and scaling them across the enterprise to keep up momentum and drive organizational impact. Cloud ML adopters are significantly less likely to slow down on AI implementation; 42% disagree or strongly disagree that they will slow down their adoption of AI technologies due to any emerging risks, compared with more than 20% of others.
- Continued high investment. The high volume of activity among cloud ML users is expected to continue, with almost every respondent in the category stating their organization plans to increase its AI investments in the future (90% in cloud ML users versus ~70% others). The quantum of increase into the next fiscal year, however, is expected to be marginally lower among cloud ML adopters, presumably due to already matured AI programs with high budgets (24% in cloud ML users versus 26% others).
Faster ROI with cloud ML use
Cloud ML users are much more likely to have a larger volume of AI initiatives, can sustain and scale enterprise adoption, and can continue with bigger budgets. These practices in themselves have the potential to uplift any AI program and generate returns for the organization. However, based on our analysis, we find that respondents with a robust cloud ML strategy fare marginally better in terms of ROI.
Furthermore, cloud ML organizations are also able to manage and meet expectations around payback timelines, typically aggressively set by top executives. However, the same cannot be said for other organizations, where almost 40% take longer than expected to achieve intended returns (compared with only 25% among cloud ML adopters).
Cloud ML users are defined as respondents who stated their organization is using cloud-enabled AI approaches, which includes: ‘AI-as-a-service’ AND (‘Data science/ machine learning platforms or frameworks’ OR ‘Automated machine learning tools (AutoML)’); n=423; 15.5%
Others are the remaining 84.5% (n=2314) of the respondents
*Note: %age of respondents stating to Agree or Strongly agree to ‘My organization is slowing its adoption of AI technologies because of the emerging risks’
It is evident that, on top of enhanced strategic decision-making, competitive advantage, and improved process efficiencies and employee productivity, those with a cloud ML strategy see improved enterprise adoption, bigger budgets, and thus better return on investment. Organizations can respond in kind by thinking through their cloud ML strategy to advance enterprise AI programs. Whether that is with cloud AI platforms, cloud ML services, or AutoML (including low-code and no-code options), success will depend on the business cases, people, and technical requirements.
For more on this topic, check out our recent research, “Time, technology, talent: The three-pronged promise of cloud ML.”
 ReportLinker, “Global Data Lakes Industry,” April 30, 2020, https://www.globenewswire.com/news-release/2020/05/01/2025940/0/en/Global-Data-Lakes-Industry.html.
 Cloud ML users are defined as respondents who stated their organization is using cloud-enabled AI approaches, which includes “AI-as-a-service” and “Data science/machine learning platforms or frameworks” or “Automated machine learning tools (AutoML)”; n=423.