Incident sheet · /infrastructure
- System
- Infrastructure & Cloud
- Filed
- Read
- 4 min · 745 words
- Severity
- Routine
Legacy Software Drains AI Resources for 78% of IT Leaders
Google Cloud research of 1,250 U.S. leaders finds 78% say legacy maintenance drains AI resources, while 76% link single-vendor estates to systemic security risk. Buyers face mounting patch burden.

Event log
- Hypothesis Group surveyed 1,250 U.S. enterprise technology decision-makers across public and private sectors for Google Cloud.
- 78% said maintaining legacy systems diverted capital, engineering talent and resources from AI innovation.
- 59% reported unreasonable time maintaining, patching, integrating and securing legacy enterprise software.
- Microsoft addressed 3,338 vulnerabilities across four Patch Tuesdays, including 1,407 in October.
- 75% said existing software limited AI benefits; 76% linked single-vendor reliance to systemic cybersecurity risk.
By Rebecca Stone
Seventy-eight percent of U.S. enterprise technology decision-makers say legacy software diverts capital, engineering talent and resources from AI innovation.
Google Cloud research used a Hypothesis Group survey of 1,250 U.S. enterprise technology decision-makers across the public and private sectors. It also found 59% spend unreasonable time maintaining, patching, integrating and securing legacy enterprise software. The survey defines legacy software as software running locally on computers and servers rather than in the cloud.
AI is marketed as a frictionless upgrade to installed enterprise suites. Google Cloud Vice President of Global Risk and Compliance Jeanette Manfra instead calls the legacy delivery layer an “innovation tax” on AI modernization. Resource-constrained government agencies can leave system administrators and security engineers trapped between that mandate and systems they must keep resilient.
How large is the patch burden?
Microsoft addressed 3,338 vulnerabilities across Patch Tuesdays in July, August, September and October, including 1,407 in October. The research also cites 1,130 vulnerabilities addressed across all of 2025 and 729 in the first six months of 2026.
The analysis credits Microsoft with fixing vulnerabilities and says rigorous patching matters. It also warns that release-then-patch assumptions no longer match an AI-driven threat environment.
Manfra said: “AI has forever changed the ‘release-then-patch’ model of software security.” Legacy estates contain decades of technical debt, spread across billions of lines of code and dozens of interconnected products and services in and out of the cloud.
AI-enabled vulnerability discovery by attackers will accelerate the pressure. IT teams face a zero-sum choice between securing old systems and building new capabilities.
Does incumbent software restrict AI choice?
Three-quarters of enterprise leaders, or 75%, report that existing enterprise software limits their ability to take full advantage of AI. Eighty percent say an incumbent’s installed base (its existing deployed product footprint) should not determine which AI an organization deploys.
Another 78% agree vendor lock-in (dependence on one provider that restricts alternatives) limits their ability to choose the best AI. Thirty-nine percent cite complex licensing, contracts or limited negotiating power as byproducts of reliance lock-in.
Seventy-six percent of technology leaders warn that a single-vendor monoculture (reliance on one supplier across the stack) elevates systemic cybersecurity risk. Many organizations tolerate this pattern because it is not an architectural choice.
What should buyers demand?
- Restore software, cloud and AI choice. Policymakers and enterprises should require customers to choose products that fit their businesses. Incumbent legacy software running servers, email and operating systems must not constrain that decision. Some organizations will retain legacy technology; others will stop investing in systems they consider unproductive. Either path requires a genuine choice.
- Apply “secure by design” during procurement. Secure by design means setting security requirements during engineering rather than after deployment. Manfra said government teams urged manufacturers to take radical ownership of customer security outcomes instead of treating security as an aftermarket maintenance contract. Government and enterprise buyers should hold vendors accountable for durable systems and stop the accumulation of new technical debt.
- Use AI on deployed estates. Buyers should use AI both to ship more secure products and to find and fix flaws in already deployed software. Google deploys AI to automate threat detection and basic remediation across complex multi-cloud and on-premises environments. The CodeMender autonomous code-auditing agent and AI Threat Defense security platform use AI to discover flaws, validate risk, generate fixes and support remediation before exploitation.
Which providers can lead AI?
The survey found 84% agree AI adoption makes technology-provider choice more important than ever. Seventy-eight percent say providers that dominated the desktop and cloud eras are not best positioned to lead AI.
Manfra, who spent decades in government and enterprise cybersecurity, led federal network defense at the Cybersecurity and Infrastructure Security Agency and Department of Homeland Security. She worked alongside frontline cybersecurity professionals across the public and private sectors.
She said AI can accelerate human ingenuity, automate tedious friction and unlock mission value. But that future cannot be built on software models designed 30 years ago.
As enterprise teams continue applying emergency patches to legacy systems, chief information officers and chief information security officers must test each technology partner’s direction. Manfra closed with a direct question: “Is your technology partner helping you actively build the future, or keeping you on a treadmill of tech debt running hard just to stay in place?”
via federalnewsnetwork.com (Source)
More from Rebecca Stone
Show full bio
Senior reporter covering money, funding and market shifts around enterprise information technology and the leaders who run at CIO Monitor.
208 articles
The newsroom digest
One email a day with the stories that matter. No noise.
Cross-references
- Accenture and Google Cloud Field 1,000 Engineers for Enterprise AI
- Windows Server 2028 Preview Targets On-Prem AI and Self-Healing Infrastructure
- Microsoft fixes 1,235 flaws, one exploited Outlook zero-day
- Red Hat to staff Singapore AI hub with 18 as patching lags
- Microsoft Commits $2.5 Billion to New AI Deployment Company