Trends
Nadel Phelan continuously curates insightful industry trends. To receive our monthly trend report, click here.
Artificial Intelligence • March 2026
Agentic AI expands the attack surface; security shifts from “model safety” to “agent control”
As enterprises embed AI agents into workflows, the dominant risk moves from prompt-quality to tool access, identity/privilege abuse, and goal hijacking—i.e., an agent doing the wrong thing “legitimately” with the permissions it has. OWASP’s Agentic Applications Top 10 (2026) frames this as a distinct risk class (tool misuse, delegated identity abuse, data exfiltration via actions). The 2026 implication: treat agents like a new tier of software principals—least-privilege tool scoping, action approval gates, audit logs, and non-human identity controls become baseline requirements.
Artifical Intelligence • March 2026
AI infrastructure capex accelerates; “compute + power” becomes strategic risk
AI spending is increasingly driven by infrastructure buildout rather than experiments. Recent reporting projects major hyperscalers investing on the order of $650B in AI infrastructure in 2026 (up sharply from 2025), with growing dependency on power, GPUs, and data-center supply chains. The 2026 implication: AI roadmaps will be constrained less by model ideas and more by capacity planning, cost governance, and resilience (power pricing/availability, vendor concentration, regional outages).
Cybersecurity • March 2026
US Agencies Warn of Iran-Linked Cyber Threats to Critical Sectors
National security agencies previously issued joint advisories that Iran-affiliated cyber actors may target vulnerable U.S. networks and “entities of interest,” including critical infrastructure operators. Threats include password spraying, MFA manipulation, reconnaissance, and hack-and-leak operations. Continued vigilance and hardening of exposed services are recommended.
Cybersecurity • March 2026
Ongoing Vulnerability Advisories & Patch Priorities
Weekly threat roundups continue to list significant CVEs affecting a wide range of enterprise and open-source software — from Windows, Android, and cloud platforms to frameworks like Apache Struts and OpenVPN. Security teams should prioritize patching high-impact exposures and integrate continuous vulnerability scanning into operational risk programs.
Cybersecurity • March 2026
Cyber Domain Escalates Amid Iran–Israel–US Tensions
Following recent strikes on Iranian targets, Iranian digital infrastructure experienced widespread disruption and attacks on civilian apps and state services — indicating cyber operations coinciding with kinetic conflict. Analysts warn this marks a new escalation that could broaden into retaliatory cyber activity against U.S. & allied systems (critical infrastructure, OT networks, water/energy utilities).
Cybersecurity • March 2026
AI-Driven Attacks Accelerate Across Enterprise Networks
Recent industry reporting highlights that AI-enabled attacks are no longer theoretical — they’re operational. Attackers are using generative AI and automation to conduct faster reconnaissance, credential theft, and lateral movement, with average breach breakout times dropping to ~29 min in 2025 (65 % faster than 2024). AI prompt injection and malicious AI endpoints are emerging as new attack vectors. CrowdStrike urges security teams to build rapid response and detection capabilities that match the attackers’ speed (credential theft, ransomware, fake AI servers).
Discussing AI and IP
Passage to Profit
Exploring AI’s Impact on Business
Tech Talks Daily
Exploring AI’s Impact on Marketing
SkyeLine
Studio
Just for Fun
Puzzle
Artificial Intelligence
Where AI and Humans Meet – A Tale of Collaboration
For me, collaboration works when I can anticipate the questions an acquirer will ask, not just from individuals who have something to sell but from the buyer’s organization as a whole.
How will this integrate with their culture, budgets, staffing, networks, facilities, processes, and applications? What are the points of commonality? Where are the points of disagreement? When I can anticipate those questions, when I know the issues before they’re voiced, that is when I know collaboration is working.
Artificial intelligence has changed this process in a critical way. A deep analysis that once consumed a month of effort recently took me seven hours. That kind of acceleration changes everything, not just for me but for anyone willing to collaborate with AI.
Of course, AI has its limitations. Everyone has heard the horror stories of hallucinations, sloppy outputs, even stray prompts left in. I would not turn it loose in an enterprise without significant guardrails. But when AI is treated as a true collaborator, with tight control and continuous oversight, it becomes extraordinary.
The key is not to hand over the work wholesale but to collaborate in pieces. Read every sentence as it comes back. Guide the conversation. Push deeper. That is where experience matters: knowing when you have reached a real stopping point and when there is a 100 feet left to go.
One of the enduring surprises of my career came from focus groups and surveys. The results often revealed insights I never anticipated. AI can do the same. It will push back, challenge my logic, and force me to go further. I have set my GPT to always let me know if there is more relevant information to consider.
That is what makes AI a true collaborator. It helps me find the middle ground, the common ground, faster and with greater confidence. And in the end, that is what matters most in fundraising and M&A: bringing people together around a story they can all believe in.