AI spending surges, but governance gaps stall real progress

By early 2026, a strong commitment to AI investment was evident, with 74% of global leaders prioritizing AI spending despite economic uncertainties. However, KPMG research revealed that hefty investments alone failed to guarantee clear operational value or ROI, as many organizations remained hesitant to fully embrace AI beyond initial steps. This cautious stance persisted even as 32% of firms began deploying agentic AI at scale, indicating that while technological advances progressed, foundational challenges from earlier AI phases continued to hinder true value realization.

A widening gap emerged between corporate confidence and actual AI success, particularly in the U.S., where 76% of companies believed they led in AI adoption but only 10% qualified as true 'AI Leaders' making significant progress and realizing ROI. This overestimation stemmed largely from conflating isolated pilots with enterprise-wide integration, as nearly two-thirds remained stuck in experimentation phases. EXL’s EVP Anand Logani emphasized that the core barrier was not technology but the failure to transform operating models, with only 44% of leaders having redesigned processes to embed AI effectively compared to 23% of laggards.

In Canada, nearly half of business leaders were trapped in AI experimentation without meaningful ROI, underscoring a global investment-value disconnect. BDO Canada’s Bill Syrros highlighted that many executives lacked clarity on accountability, data reliability, and value measurement, widening the divide between those who understand AI’s practical implementation and those who do not. Compounding this, 27% of Canadian leaders underestimated AI’s future impact, suggesting a visibility gap that threatens to stall progress unless organizations adopt clear governance, workforce enablement, and outcome-focused AI strategies.

By mid-2026, executives warned that rapid AI adoption by staff was outpacing the readiness of security and IT teams, creating operational risks that investment enthusiasm alone could not mitigate. Kevin Hanes of Quorum Cyber cautioned about this governance lag, while Ashutosh Garg of Eightfold AI noted that despite early productivity gains reported by employees, fundamental changes in work design remained rare. Smaller firms faced an 'AI applicability barrier,' shifting focus from experimentation to scrutinizing cost, governance, and security as AI spending and token usage surged, highlighting that without robust frameworks, scaling AI safely and effectively remains elusive.

By mid-2026, a glaring disconnect emerged between soaring AI investments and the establishment of robust governance frameworks, with Galorath's report revealing that while 79.1% of organizations increased AI spending, only 23.6% had clear, documented AI policies. This governance gap fuels widespread shadow AI usage—unregulated AI adoption occurring in 70.9% of organizations—which not only undermines trust and compliance but also creates porous oversight environments. Charles Orlando highlights that organizations achieving a 51% improvement in planning accuracy are those that pair AI adoption with centralized governance and process redesign, underscoring governance as a critical lever for realizing AI’s operational value.

The absence of comprehensive AI governance and data permission frameworks has become a critical bottleneck, with Transcend Research reporting that 81% of enterprises have delayed, scaled back, or abandoned AI initiatives due to these gaps. Only 15% of enterprises possess full AI governance capabilities, and legacy consent systems fail to dynamically enforce data permissions across AI workflows, creating architectural vulnerabilities that erode trust and hinder sustainable scaling. Encoded AI Governance, which automates data use rule enforcement within systems, is proposed as a promising solution to close these governance and security gaps.

Sector-specific analyses, such as Grant Thornton’s study on multifamily firms, reveal that rapid, vendor-driven AI adoption without adequate organizational controls exacerbates governance risks, with only 13% of leaders confident in passing an AI governance audit. Firms that intentionally invest in governance foundations, clear ownership, and measurement definitions—not merely rapid adoption—are the ones realizing measurable AI returns. This pattern echoes BDO’s findings in Canada, where insufficient governance and workforce readiness stall AI scaling and limit integration into operational workflows, highlighting governance as a linchpin for trust, compliance, and value realization.

Emerging AI-specific security threats, such as prompt injection attacks and shadow AI, expose critical vulnerabilities as 65% of organizations lack dedicated defenses and over half of AI-driven work occurs outside official systems, creating compliance blind spots. Traditional security teams are ill-equipped to handle these novel attack surfaces, prompting calls from experts like CrowdStrike and Quorum Cyber’s Kevin Hanes for AI-native defenses and external expertise. Furthermore, rapid AI adoption outpaces governance and security preparedness, with firms grappling with cost visibility, auditability, vendor lock-in, and national security concerns, underscoring that sustainable AI scaling demands integrated governance, infrastructure, and security frameworks rather than isolated technology deployments.

By early 2026, sectors like real estate highlighted the foundational challenge of messy, unstructured data and a pervasive lack of data management skills, with brokerages often ill-equipped to organize information effectively. As one expert noted, "most brokerages aren't good at that," underscoring how poor data literacy compounds AI implementation difficulties. Yet, AI’s promise to automate interpretation and reduce training demands offers a glimmer of hope, enabling some to bypass traditional data skill gaps through what was described as "magic" integration techniques.

Data permission and governance gaps have emerged as a critical bottleneck, with a Transcend Research report revealing that 81% of enterprises delayed or abandoned AI projects due to inadequate infrastructure. Legacy consent systems, unable to dynamically enforce permissions, cripple AI’s scalability and revenue potential, prompting calls for solutions like Encoded AI Governance to automate data use rules and close this gap. This governance deficit is echoed across industries, where fragmented ownership and regulatory compliance complexities further stall AI progress.

Persistent data fragmentation and siloed systems remain the Achilles’ heel of AI scalability, as evidenced by EXL’s finding that 70% of organizations cite data infrastructure as their top barrier. While AI leaders have made strides—44% achieving enterprise-wide data access compared to just 17% of laggards—many still grapple with privacy concerns, limited model transparency, and integration friction with legacy ERP, CRM, and HR systems. This fragmentation not only undermines reliable AI outcomes but also inflates operational inefficiencies, such as the 15-30% energy waste in commercial buildings due to disconnected systems.

A growing consensus among IT leaders, particularly in regions like India and Southeast Asia, is that AI’s true bottleneck lies not in the technology itself but in the underlying data infrastructure. With 79% of Indian IT leaders citing poor real-time data pipelines as a major hurdle, and 91% recognizing data streaming platforms as essential for trustworthy, contextualized data, investments are increasingly shifting toward integrated, API-first architectures. As Confluent’s Shaun Clowes put it, "Most organisations do not have an AI investment problem, they have a data problem," emphasizing that continuous intelligence demands a foundational overhaul of fragmented, batch-oriented systems.

By mid-2026, it became clear that moving AI from isolated pilots to enterprise-wide integration demands a fundamental transformation of operating models, workflows, and governance structures. Studies show that 44% of AI leaders have completely redesigned their operating models to embed AI deeply, compared to only 23% of laggards, underscoring the competitive gap. Anand “Andy” Logani emphasized that successful AI execution requires reimagining decision-making processes as if AI were built in from the start, not just investing in technology. This holistic transformation involves establishing cross-functional governance boards to address model transparency, bias, and compliance, while also redefining talent roles and team structures to align technology tightly with business strategy.

Reports from BDO Canada and FPT in late June and early July 2026 highlight that most organizations remain stuck in AI experimentation due to insufficient operational readiness, with only 18-39% making meaningful progress in embedding AI into workflows or aligning strategy, governance, and operating models. Bill Syrros of BDO Canada warns that the real divide will be between organizations redesigning work around AI and those funding disconnected pilots. Effective scaling requires clear governance, workforce enablement, adoption planning, and outcome measurement tied directly to business KPIs, moving beyond automation and cost reduction toward an AI-first operating model.

In Southeast Asia and other regions, integration challenges with legacy ERP, CRM, and HR systems, alongside a critical shortage of MLOps engineers, create significant friction in embedding AI deeply. Enterprises are adopting phased approaches—starting with data quality audits, then scoped deployments on repetitive tasks, and culminating in organizational change—to overcome these hurdles. This necessitates deliberate investment in API-first architectures and middleware layers, as well as reliance on external vendor partnerships to bridge talent gaps, illustrating that operating model redesign must encompass both technological and workforce capability transformations.

Sector-specific insights from construction, multifamily housing, and commercial real estate reveal that embedding AI requires not only new operating procedures and digitization to generate structured data but also a mindset shift around budgeting for AI-related skilled resources, which can constitute 20-30% of costs annually. For example, EliseAI reports only 34% of operators have fully embedded AI into daily operations despite 89% having introduced it, with integration challenges and legacy systems cited as major barriers. Furthermore, workforce shortages—such as 43% understaffing in US facility teams per CBRE—intensify the need for operational model transformation to leverage AI-driven efficiencies, emphasizing that AI must become the core infrastructure rather than an add-on.

The real estate sector, especially residential brokerage and multifamily firms, grapples with deeply unstructured and messy data that severely impedes AI deployment. As one analysis from April 2026 highlights, brokerages often lack the capability to clean and organize their data effectively, a challenge compounded by poor training practices that translate directly into AI shortcomings. Yet, those who persevere in compressing through this data chaos gain a lasting competitive advantage, with Grant Thornton's June 2026 survey underscoring that multifamily firms achieving measurable AI returns are those investing deliberately in governance, ownership clarity, and measurement frameworks rather than rushing adoption.

Multifamily real estate is pivoting AI investments from back-office automation toward front-office applications like leasing, resident engagement, and portfolio analytics, which directly impact KPIs such as lead response time and resident satisfaction. However, realizing AI’s full value hinges on clean data integration and workflow automation that enable faster, informed decision-making and free property teams to focus on value-added activities. Despite 89% of operators introducing AI, only 34% have fully embedded it into daily operations, with integration challenges, legacy technology, and vendor fragmentation cited as persistent barriers—issues that 76% of operators face while juggling multiple AI vendors.

Industrial maintenance stands out as a sector rapidly embracing AI, with 58% adoption and 75% of users reporting ROI within six months, driven by integration of AI with CMMS platforms and IoT sensors to create real-time digital maintenance ecosystems. This enables predictive fault diagnosis and risk-based inspections that reduce catastrophic failures and optimize asset integrity. Yet, success depends equally on overcoming operational hurdles such as skilled labor shortages, legacy system integration, and cultural resistance, underscoring that AI’s promise is as much about preparing human teams and redesigning workflows as it is about technology deployment.

Construction faces a unique AI adoption challenge rooted in limited software budgets and the absence of structured data systems and operating procedures, making the establishment of systems of record a critical priority. As Nvidia’s Jensen noted in mid-2026, AI usage can consume 20-30% of skilled resource costs, a significant financial consideration for business owners. However, digitization combined with AI offers a transformative path from a fragmented, reactive industry toward a collaborative, data-driven environment that accelerates project delivery and enhances risk management. This evolution is further enabled by the democratization of AI interfaces—from complex mathematical models to accessible language-based tools like ChatGPT—broadening adoption across real estate and construction workflows.

Commercial real estate confronts operational complexity from fragmented, siloed systems that hinder AI-driven operational intelligence and energy efficiency, with research showing 15-30% energy waste due to system faults and poor controls. Facility teams struggle with multiple vendor platforms and dashboards, slowing decision-making and creating operational drag, while only about a third of organizations report maturity in digital integration or AI scaling. Yet, projects like the Pennsylvania Convention Center demonstrate that operational modernization through AI and digital infrastructure can yield substantial financial and sustainability benefits—18% energy reduction and nearly $700,000 in savings—highlighting the sector’s urgent need to build connected data infrastructures to make intelligent, portfolio-wide operations a competitive asset.

Without targeted upskilling and clear ownership, even the best AI strategy flounders as teams lack the readiness to operationalize new technologies.