Microsoft Customers Rely On Azure For AI-Readiness To Adopt And Run GenAI Security In The Cloud

According to Forrester Research, generative AI (genAI) had made AI capabilities more than an IT project — it’s now an everyday productivity tool. Easy access to AI and the rapid consumerization, adoption, and possibilities of ready-to-use genAI apps has captured everyone’s interest. In the past, AI projects concentrated on internal use cases meant to drive process improvements and cost reductions for the organization. Today, AI leaders are tasked with delivering upside and demonstrating how AI is contributing to growth and company value in addition to impacting the bottom line.¹ Therefore, organizations must accelerate AI adoption and scale ahead of the competition to effectively impact business performance. Organizations increasingly look to the cloud to circumvent cost, resource, and performance limitations of prior environments and ramp genAI development initiatives securely.

Microsoft commissioned Forrester Consulting to evaluate how organizations are using Azure for AI-readiness to adopt and run genAI securely in the cloud. Forrester conducted a survey of 112 representatives from organizations using Azure for AI-readiness and aggregated data from 10 related Total Economic Impact™ (TEI) studies to explore this topic.² This analysis evaluates the risks associated with adopting genAI in the cloud and how Microsoft Azure enables organizations to scale AI projects in a cost-efficient, performant, and secure way. It will focus specifically on organizations that use Microsoft Azure as their infrastructure to deploy genAI in the cloud. To learn more about the data sources for this study, see Appendix A.

The survey respondents noted their organizations wanted to use genAI to deliver better end-user experiences that differentiate their business and accelerate growth, such as enhancing end-user productivity when interacting with the organization. They also wanted to target internal use cases like automating processes to drive productivity for data scientists, genAI engineers, and non-IT business employees.

The respondents said their organizations adopted Azure to promote a culture of innovation that included taking advantage of AI technology with more flexibility and less risk than prior on-premises deployments. In fact, a survey conducted in “The Total Economic Impact Of Migrating To Microsoft Azure For AI-Readiness” found that 75% of survey respondents whose organization migrated to Azure for AI-readiness reported that the migration was necessary or significantly reduced the barriers to enabling AI/machine learning (ML) at their organization.³ In the most recent survey conducted for this analysis, 65% of the 112 respondents agreed that deploying genAI in the cloud would help meet organizational objectives to avoid technology restrictions and the limitations of legacy on-premises deployments.⁴

However, once organizations were in the cloud, they faced challenges in safely scaling their AI deployments. Surveyed IT and security decision-makers cited security concerns as their primary challenge in deploying genAI in the cloud. More than 50% of respondents saw insufficient knowledge about genAI security risks; data privacy and security issues from proliferation of genAI applications and content; and a lack of developer expertise as barriers to adopting genAI in the cloud. Additionally, some survey respondents cited lingering challenges associated with on-premises infrastructure, including weak existing functionality and integration issues (30%), high costs associated with supporting genAI apps on-premises (23%), and holdover delays and expenses related to migration projects (27%).

The lack of in-house knowledge and expertise and persisting challenges associated with legacy infrastructure further contributed to challenges identified in prior TEI studies, such as slower development efforts and poorly built custom models that are, at best, labor-intensive or costly to scale and, at worst, not properly governed or ethical.⁵

Survey respondents reported that their organizations partnered with Azure to combat these challenges and meet organizational goals for genAI use cases. GenAI applications and content are only as good as their underlying data. To that end, respondents appreciated Azure’s integration and data management capabilities, including their colocation strategy that eliminates data silos and optimizes performance at scale to meet genAI objectives, listing it as the top reason for partnering with Azure for deploying genAI (595). Other top reasons include cloud Azure’s security capabilities (58%) and genAI transparency and accountability (50%) (see Figure 1).

Azure cloud security and compliance capabilities begin at the infrastructure and platform level with the use of products such as Defender for AI and/or for Cloud Apps, Microsoft Purview, Data Security and Compliance for M365 E5, Sentinel, Azure Key Vault, and Entra ID. GenAI workloads running on the Azure infrastructure benefit from the same protection across the lifecycle from development to runtime and beyond, protecting key areas such as data privacy concerns through Azure Key Vault; development lifecycle protection through Defender; security threat information and event management through Sentinel; privacy and access control through Entra; and data governance protection through Purview. Over 50% of survey respondents identified Defender for AI, Defender for Cloud Apps, Sentinel, and/or Azure Key Vault as the Azure cloud security and compliance capabilities that their organization finds most important for delivering genAI success.

Additionally, a combination of Azure products (including Azure Virtual Machines, Azure VMware Solutions, Azure AI Infrastructure, Azure App Services, Azure SQL Managed Instance, Azure SQL Database, Azure Cosmos DB, Azure AI Services, Azure ML Services, and Azure Database for PostgreSQL) help customers build, run, and manage applications and provide purpose-built infrastructure to scale genAI transformation with improved stability and reduced cost. These tools work together to ensure that genAI applications and content are built soundly and in compliance with security and governance standards meant to protect the organization, the developers, and the end users.

Al can be scary but Azure users find solace in using tools and technology on a cloud infrastructure that they already trust from a security standpoint. AI services and tools accessible through Azure enable organizations to build genAI securely and at the scale required to meet business needs for both internal and external use cases.

INVESTMENT DRIVERS For DEPLOYING GENAI WITH AZURE

Respondents noted their organizations adopted Azure to deploy genAI in the cloud at scale while navigating a complex security and compliance environment. These organizations struggled with several challenges related to their legacy environments or perceived risks of adopting genAI in the cloud, including:

• Performance limitations and the high cost of on-premises infrastructure. Barriers to success with genAI included the high costs of managing legacy infrastructure as well as resource time dedicated to infrastructure management that took time and dollars away from supporting key business strategies.⁶ Additionally, on-premises environments were saddled with outdated security solutions, such as on-premises security information and event management (SIEM) solutions that were expensive, required specialized staff, and failed to meet evolving security and compliance needs.⁷ Moreover, legacy environments were housed data in siloed systems which made it more difficult to pull that data together for genAI content and applications. Managing the underlying data in this fragmented way often resulted in poor performance. Overall, organizations were behind on meeting their organizational genAI objectives.
• Security concerns. Security remained the top concern among survey respondents when it came to deploying genAI in the cloud. Many security concerns focused on the unknown security landscape of genAI. For example, a lack of knowledge about genAI best practices and controls contributed to higher risk. As did the potential of building models that were not properly governed or compliant — especially given added data privacy concerns associated with genAI content and increasing regulatory liability in the space. Survey respondents also surfaced security concerns related to general cloud adoption, such as growing cloud consumption risk and lingering security risks associated with legacy environments.
• Resource capacity. Where security concerns were not obviously highlighted, they still lurked below the surface of the remaining challenges cited in the survey results. For instance, 57% of survey respondents cited resource capacity restrictions, either due to the impact of managing legacy infrastructure or from pressing organizational priorities, as a barrier to genAI adoption in the cloud. When pressed on the outcomes of these capacity restrictions, respondents were worried about the additional security concerns brought on by dangerous workarounds, such as the use of IT systems, devices, software, or applications outside of the organization’s centralized IT department, also known as shadow IT (45% to 60% of respondents). Respondents also had concerns about the further risk of developing applications with genAI that did not meet strict security and compliance requirements (54% to 55% of respondents).⁸

Features THAT SUPPORT GENAI INITIATIVES

Respondents’ organizations chose to invest in Microsoft Azure for its secure environment and enhanced AI functionality that, together, propelled secure innovation with AI at the enterprise level. Respondents cited specific features, including (see Figure 2):

• Azure security features. Fifty-eight percent of survey respondents indicated that their organization selected Azure to build and deploy genAI in the cloud specifically for its integrated cloud security and compliance capabilities that enabled them to be genAI-ready more quickly and securely. Those capabilities included:
  • Microsoft Sentinel. Fifty-two percent of respondents found Microsoft Sentinel to be the most important Azure security and compliance capability when it came to genAI success. Microsoft Sentinel is a cloud-based solution that enables organizations to eliminate extraneous security infrastructure and operational costs while scaling to meet organization demands, including those meant to scale genAI use cases.⁹
  • Defender for AI. Fifty-two percent cited Defender for AI and another 50% cited Defender for Cloud apps as the most important Azure capabilities for genAI success. Microsoft Defender secures workloads in the cloud by reducing attack surfaces and protecting especially vulnerable workloads, such as those associated with continuous integration and delivery lifecycles. As a result, survey respondents found that they could reduce risk and improve security and compliance postures at scale and maintain these postures in dynamically shifting regulatory climates, such as those currently associated with genAI.¹⁰
  • Microsoft IaaS. Additionally, Microsoft Azure infrastructure could be managed as a service through Microsoft information as a service (IaaS). IaaS is positioned as an alternative to self-managing, traditional on-premises infrastructure for organizations with the objective of refocusing resources on strategic needs of the business, such as powering AI initiatives.¹¹ Microsoft IaaS also improved security in the face of increasing cyberattacks and supported key security strategies and compliance with regulations at a global scale for sensitive, regulated workloads, such as those that include genAI.¹²

• Azure AI functionality. Thirty-seven percent of survey respondents cited Azure’s advanced AI and ML tooling as the reason for selecting Azure for building and deploying genAI in the cloud, noting best-in-class functionality and ease of use as well as cost savings afforded from vendor consolidation. These tools included (see Figure 3):
  • Azure AI services. Fifty-eight percent of survey respondents said their organization uses Azure AI services. A prior Forrester TEI study titled “The Total Economic Impact Of Microsoft Azure AI” found that Azure AI services increased the quality and scalability of AI and ML models at interviewees’ organizations. Enhanced model compliance and governance and assurance of responsible AI/ML practices provided the necessary transparency to reduce error rates in model outputs. Azure AI services helped the interviewees’ organizations quantify model fairness, robustness, and security to ensure that their models were properly governed and adhered to ethical principles¹³. Additionally, some interviewees utilized Azure AI services to fuel genAI use cases that helped identify and manage risk and reduce the risk of compliance violations, penalties, and fines meant to enhance accountability in the space.¹⁴
  • Azure OpenAI. Sixty-eight percent of survey respondents identified Azure OpenAI as the most important Azure AI or ML tool to their success with genAI. Respondents’ organizations utilized Microsoft Azure OpenAI services from the CoPilot stack to integrate OpenAI models into their applications that were supported by Microsoft’s enterprise-grade commitments to ensure AI privacy, safety, and security. In this way, respondents were able to meet business objectives by improving end-customer experiences, such as generating revenue, without the added security concerns associated with OpenAI.
  • Retrieval-Augmented Generation (RAG). Fifty-one percent of survey respondents identified RAG as the most important Azure AI/ML tool to their genAI success. RAG is an architecture that augments capabilities of large language models, such as ChatGPT, by adding an information retrieval system that provides grounding data to exercise more control over formulated responses and more constraint over embedded models to improve both accuracy and security.¹⁵

• Azure data management and colocation strategy. Fifty-nine percent of survey respondents cited data integration and management capabilities through Microsoft Fabric, Azure Synapse, and Azure Databricks as the reason their organization selected Azure for building and deploying genAI in the cloud. Before deploying genAI workloads with Azure, legacy on-premises environments restricted genAI success because of existing technology limitations, including poor integration capabilities and a fragmented approach to data management. As a result, genAI workloads became costly to deploy and manage and performed poorly. Thirty-one percent of survey respondents cited Azure’s colocation strategy specifically to circumvent these challenges and drive better genAI performance at scale.

Key Results

The results of the investment include:

Redirected resources away from managing infrastructure to refocus on business growth. On-premises environments required many resources to run and manage infrastructure. Additionally, legacy infrastructure was not tuned to deploy genAI applications successfully, adding time to the development effort to build and deploy. Deploying genAI in the cloud with Azure enables organizations to redistribute resources to more value-add work, such as time spent on genAI initiatives. As a result, organizations can:

• Refocus resource time on innovative initiatives, including genAI. A prior study titled, “The Total Economic Impact Of Microsoft Azure IaaS” found that survey respondents used Microsoft IaaS to manage their cloud infrastructure indicated that 80% of infrastructure service professionals were redirected to more strategic initiatives, such as cloud migration projects meant to deliver genAI use cases to the business at scale.¹⁶ Additionally, Azure brings modern security tools to the table, including Microsoft Sentinel. Decision-makers who were interviewed in a prior study titled “The Total Economic Impact Of Microsoft Sentinel” whose organizations use Azure suggested that 16% of legacy SIEM specialists were reallocated to value-added initiatives, such as securely developing and deploying genAI applications.¹⁷  
• Enable more productive application development efforts. Respondents noted Azure AI services improved productivity for data scientists, AI/ML engineers, and employees whose work is impacted by AI. Based on a previous TEI study, we found that utilized Azure AI services redirected up to six FTEs to take on more complex development projects.¹⁸ Additionally, Microsoft Defender reduced development efforts and improved time to value to develop secure products that enabled business growth by providing a 30% productivity improvement for SecOps teams due to expanded visibility, context, and automations. It also did so by reducing the volume of false positives by 50% and decreasing the time spent on investigation and remediation efforts by 30%. The time savings freed up SecOps teams to focus on more proactive threat-hunting and other higher-value activities, such as securing and monitoring genAI workloads in the cloud.¹⁹

Improved security posture and ability to meet compliance standards. Our studies have found that Azure improved security postures at the infrastructure, platform, and workload or application level through tooling and inherent capabilities. Additionally, a combination of tooling, transparency, and time savings made it easier for organizations to meet compliance standards and regulations. As a result, organizations can:

• Reduce the risk of a breach and recover associated costs. In “The Total Economic Impact Of Microsoft Sentinel,” interviewees indicated that Sentinel reduced the risk of a breach by 35% from more visibility into the risk profile.²⁰ Additionally, another TEI study found that Microsoft Defender reduced the risk of a breach by 10% by detecting more incidents that would have previously gone undetected.²¹ A reduction in risk of a data breach translated into avoided potential costs, such as those associated with brand and reputation impact, downtime impacts for internal employees, or missed revenue opportunities.
• Improve compliance posture to avoid costs associated with additional overhead. Because governance is commonly perceived as a bottleneck to innovation, it’s tempting to set it aside and move quickly on AI in response to pressure from the C-suite and individual use. However, many machine learning models are opaque, making it difficult to interpret and control model behavior. With genAI, there’s the added challenge of hallucinations that look reliable but are incorrect.²² In a TEI study focused on Microsoft Defender, we found that Microsoft Defender helped interviewees’ organizations meet governance requirements for their existing workloads in the cloud without costly overhead or error-prone processes.²³ This TEI also noted that workloads protected by Defender for Cloud reduced audit compliance overhead by 15% annually.²⁴ At the infrastructure level, Azure IaaS supported key security strategies and compliance with regulations across geographies such that organizations entrusted sensitive, regulated workloads to Azure IaaS.²⁵ Organizations further reduced compliance overhead with Microsoft Sentinel by streamlining compliance reporting and avoiding spend on external consultants.

Improved throughput/performance of genAI applications. Building and deploying genAI applications in the cloud with Azure improved the performance of those applications by enhancing the management and integration strategies for the underlying data. Additionally, more resources were free to upskill in AI and focus on innovative development projects to meet strategic genAI initiatives. As a result, organizations:

• Benefited from facilitated innovation. Survey respondents were asked about their flexibility to build and improve AI and ML applications and their ability to innovate. Respondents with Azure responded confidently at more than twice the rate of those with on-premises infrastructure. Respondents cited that Azure offered increased flexibility to build and change AI applications with lower risk and the ability to reinvest in resources previously focused on infrastructure and upskill them in AI for the vote of confidence.²⁶
• Improved application performance. With more resources spending time upskilling in AI best practices and a better underlying infrastructure to support genAI development efforts, respondents noted their organizations experienced better application performance. Forty-one percent of survey respondents cited better throughput for genAI apps and content running on Azure compared to other vendors. Additionally, a prior TEI focused on Azure AI services indicated that organizations experienced better quality and scalability of AI and ML models and fewer error rates in model outputs.²⁷ As a result, organizations improved end-user experiences.  
• Met business goals associated with genAI. As noted, survey respondents had many goals for genAI applications that spanned both internal business and external end-user use cases. The cost and resource savings enabled by moving to the cloud made it easier for organizations to focus on scaling AI initiatives to meet these goals.²⁸ Previous studies measured the impact of meeting business goals associated with AI use cases. For example, the TEI of Azure OpenAI services recorded business growth from higher conversion rates and leads, better customer retention, or higher revenue per customer as measured as a 6.85% improvement to conversion rate for prospects. The TEI of Azure AI measured a total reduction in costs by 7% attributed to ML models that were used to optimize costs across the business. Another use case from the same TEI cited cognitive services and applied AI services to support customer-facing services or revenue streams that also contributed to revenue growth as measured by a 150% increase in work output.

Appendix A: SUPPLEMENTAL MATERIAL

For purposes of this spotlight, Forrester reviewed details from seven previous TEI studies and surveyed 112 representatives of organizations that currently use Azure.

Total Economic Impact studies included in this analysis include:

• “The Total Economic Impact™ Of Migrating to Microsoft Azure for AI-Readiness,” June 2024.
• “The Total Economic Impact™ Of Microsoft Azure AI,” April 2023.
• “The Total Economic Impact™ Of Microsoft Defender for Cloud,” August 2024.
• “The Total Economic Impact™ Of Microsoft Security,” February 2023.
• “The Total Economic Impact™ Of Microsoft Azure Resilience Guidance,” June 2024.
• “The Total Economic Impact™ Of Microsoft Azure IaaS,” June 2023.
• “The Total Economic Impact™ Of Microsoft Sentinel,” December 2023

Survey respondent demographics include:

• Surveyed organizations ranged in size, with over 80% of representatives indicating that their organizations had between 1,000 and 14,999 employees and generated between $500M-$9.99B in annual revenue.
• Respondents represented organizations across industry segments, including Retail (11%), Manufacturing (9%), Technology (8%), and Financial Services (7%).
• All respondents were manager-level or above in IT/Technology (44%) or Security, Compliance, Operations (56%) roles, and, at the very least, influenced decisions regarding security, compliance, identity, endpoint management, and privacy for their organization.
• 72% of respondents were using Azure to support GenAI deployments for less than 1 year and over 63% of respondents had successfully deployed GenAI in the cloud.
• While all respondents were current Azure customers, 64% indicated that they used or planned to use a hybrid infrastructure, including a mix of on-premises and cloud, to deploy GenAI.

Survey Demographics

*Note: Percentages may not total 100 due to rounding

Appendix B: Endnotes

¹ Source: Align AI Strategy Across Three Communities To Grow AI Value, Forrester Research, Inc., September 26, 2024.

² Total Economic Impact is a methodology developed by Forrester Research that enhances a company’s technology decision-making processes and assists vendors in communicating the value proposition of their products and services to clients. The TEI methodology helps companies demonstrate, justify, and realize the tangible value of IT initiatives to both senior management and other key business stakeholders.

³ Source: Align AI Strategy Across Three Communities To Grow AI Value, Forrester Research, Inc., September 26, 2024. Source: “The Total Economic Impact™ Of Migrating To Microsoft Azure For AI-Readiness,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2024.

⁴ Base: 112 decision-makers from organizations that use Azure for AI-readiness; source: A commissioned study conducted by Forrester Consulting on behalf of Microsoft, October 2024

⁵ Source: “The Total Economic Impact™ Of Microsoft Azure AI,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, April 2023.

⁶ Source: “The Total Economic Impact™ Of Migrating To Microsoft Azure For AI-Readiness,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2024.

⁷ Source: “The Total Economic Impact™ Of Microsoft Sentinel,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, December 2023

⁸ Base: 112 decision-makers from organizations that use Azure for AI-readiness; source: A commissioned study conducted by Forrester Consulting on behalf of Microsoft, October 2024

⁹ Source: “The Total Economic Impact™ Of Microsoft Sentinel,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, December 2023

¹⁰ Source: “The Total Economic Impact™ Of Microsoft Defender For Cloud,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft August 2024.

¹¹ Source: “The Total Economic Impact™ Of Microsoft Azure IaaS,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2023.

¹² Source: “The Total Economic Impact™ Of Microsoft Azure IaaS,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2023.

¹³ Source: “The Total Economic Impact™ Of Microsoft Azure AI,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, April 2023.

¹⁴ Source: “The Total Economic Impact™ Of Microsoft Azure AI,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, April 2023.

¹⁵ Source: Retrieval Augmented Generation (RAG) in Azure AI Search, Microsoft, September 3, 2024.

¹⁶ Source: “The Total Economic Impact™ Of Microsoft Azure IaaS,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2023.

¹⁷ Source: “The Total Economic Impact™ Of Microsoft Sentinel,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, December 2023

¹⁸ Source: “The Total Economic Impact™ Of Microsoft Azure AI,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, April 2023.

¹⁹ Source: “The Total Economic Impact™ Of Microsoft Defender For Cloud,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft August 2024.

²⁰ Source: “The Total Economic Impact™ Of Microsoft Sentinel,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, December 2023

²¹ Source: “The Total Economic Impact™ Of Microsoft Defender For Cloud,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft August 2024.

²² Source: Align AI Strategy Across Three Communities To Grow AI Value, Forrester Research, Inc., September 26, 2024.

²³ Source: “The Total Economic Impact™ Of Microsoft Defender For Cloud,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft August 2024.

²⁴ Source: “The Total Economic Impact™ Of Microsoft Defender For Cloud,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft August 2024.

²⁵ Source: “The Total Economic Impact™ Of Microsoft Azure IaaS,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2023.

²⁶ Source: “The Total Economic Impact™ Of Migrating to Microsoft Azure For AI-Readiness,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2024.

²⁷ Source: “The Total Economic Impact™ Of Microsoft Azure AI,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, April 2023.

²⁸ Source: “The Total Economic Impact™ Of Migrating To Microsoft Azure For AI-Readiness,” a commissioned study conducted by Forrester Consulting on behalf of Microsoft, June 2024.