Why Enterprise AI Has a Leadership Problem

April 10, 2026 · Episode Links & Takeaways

Author

AI Daily Brief
April 10, 2026

HEADLINES

HEADLINES

The SaaSapocalypse Is Over

The narrative of AI disruption wiping out incumbent software has faded on Wall Street, and the recovery is underway. AWS CEO Matt Garman pushed back directly at the HumanX Conference on Tuesday, calling the idea that companies would use Claude Code to replace platforms like Salesforce "overblown" — while still warning that firms protecting the status quo rather than leaning in would be in trouble. Goldman Sachs analyst Peter Oppenheimer thinks the worst is over for tech broadly, noting that last quarter was one of the weakest relative performances for tech stocks in 50 years, driven entirely by fears around infrastructure spending and AI disruption. Cybersecurity is the standout subsector where the disruption narrative was most overdone: Manthan Shah of WestBridge Capital put it plainly — "AI is going to massively increase the surface area that can be vulnerable, meaning the need for security is going to compound significantly going forward." Multiple analysts have upgraded cyber stocks, with Piper Sandler's Rob Owens calling AI "an opportunity, not a replacement threat."

Anthropic Tender Offer: Nobody Is Selling

Anthropic wrapped up their employee tender offer and barely anyone took the money. The offer was pegged to Anthropic's $380B February valuation — which already felt cheap when it closed and got cheaper as Claude Code took off. With secondary markets pricing Anthropic as high as a $600B implied valuation, employees with options struck at the January 2025 price of $60B are leaving a staggering amount on the table to bet on an IPO pop. The offer fell short of its full allocation, leaving outside investors unable to buy as much stock as they'd hoped. OpenAI's most recent tender had the same dynamic: only $6.6B of $10.3B in approved sales actually went through.

Anthropic Wins the Talent War

Two big executive poaches in a single week. Eric Boyd — 18-year Microsoft veteran who most recently ran the AI hardware and software team for Azure — is joining Anthropic as the new head of infrastructure, as the company takes a more active role in managing compute after years of outsourcing to cloud partners. This follows the announcement of a 3.5 gigawatt dedicated inference deal with Google and Broadcom. The Information also reports Anthropic is hiring an entire infrastructure team from leading cloud enterprises, not just Boyd. Separately, Anthropic grabbed Peter Bailis, Workday's CTO of less than a year, who'll work on reinforcement learning engineering. The hire sent Workday stock down 6.5% on the day — the market is reading it as a signal that Anthropic may have incumbent SaaS in its sights.

Musk vs. Altman

The Musk/OpenAI trial is approaching and an amended filing this week clarified what Musk is actually after: unwinding OpenAI's for-profit conversion, removing Altman and Brockman from the nonprofit board, and directing any monetary damages to the nonprofit rather than himself. OpenAI's response on X was blunt, calling it "a harassment campaign driven by ego, jealousy, and a desire to slow down a competitor." The jury trial begins at the end of the month.

Intel Joins Musk’sTerafab

Meanwhile, Intel is throwing in with Elon's Terafab chipmaking venture in Austin, providing design and construction support and overseeing the refactoring step. Terafab is targeting one terawatt of chips per year — which would make it the largest fab in the world. Intel CEO Lip Bu-Tan called it "a step change in how silicon logic, memory, and packaging will get built." Semiconductor analyst Patrick Moorhead's read: "Intel Foundry needs anchor customers. Musk needs a process partner."

MAIN STORY

The Excited Anxiety of Enterprise AI

Enterprise AI has been discussed a lot on this show lately — implicitly through Claude Code adoption curves, SaaS disruption fears, and the maturity maps framework. But what do the actual numbers inside companies say? Several major studies have landed recently with direct sourcing from executives and employees, and together they paint a striking portrait: a leadership class that is genuinely excited and genuinely anxious in equal measure, sitting atop organizations that are structurally unprepared for what they've already committed to.

ENTHUSIASM AND CRISIS IN AI LEADERSHIP

A16Z: What's Actually Working
Coding is the dominant use case — by an order of magnitude.
A16Z aggregated private data from leading enterprise AI startups deployed inside large corporations to assess real adoption, not just pilot activity. About 19% of the Global 2000 are live paying customers of a leading AI startup — rising to 29% of the Fortune 500. That means a signed top-down contract, a converted pilot, and a live product. On use cases, coding is dominant by an order of magnitude. Customer support is the next clearest category — A16Z argues it works well because the work was often already outsourced, interactions are time-bound and discrete, ROI is quantifiable through ticket counts and resolution rates, and critically, support doesn't require 100% AI accuracy because there's always a natural off-ramp to a human. On industries, tech is unsurprising. Legal is notable: A16Z argues it was "left behind by traditional enterprise software" because static workflow tools didn't fit unstructured, nuanced work — but AI is excellent at parsing dense text, reasoning over large volumes, and drafting responses. Healthcare similarly is responding to AI in ways it never did for traditional software, because AI can take on discrete administrative work (like medical scribes) without requiring a rip-and-replace of existing EHR systems like Epic.

KPMG: Agents Have Crossed the Threshold
From 11% in full deployment to 54% — in twelve months.
KPMG's quarterly pulse survey tracks executives at companies primarily with over $1B in revenue, giving longitudinal data quarter over quarter. Average anticipated AI spend over the next 12 months has nearly doubled — from $114M in Q1 of last year to $207M now. The agent deployment numbers are the most striking: in Q1 2025, only 11% of organizations had agents in full production. By Q2 that jumped to 33%. Now, for the first time, it's over 50% — at 54%. Within that, 40% are scaling or deploying, 6% are building multi-agent systems, and 9% are orchestrating. This agentic shift is coloring everything else on the survey. Cyber and employee misuse as the most difficult society-wide AI challenge is up from 32% to 44%. Resistance to agents among employees is more about skills gaps than job security fears, though both rate very high at 76% and 71% respectively. Interestingly, agents have changed hiring approaches for experienced employees (71%) more than entry-level (64%). And when it comes to talent skills: 71% of leaders value technical and programming ability for AI-related roles, but 83% value adaptability and continuous learning.

Writer + Workplace Intelligence: The Cultural Crisis
"The tension has evolved into something much more consequential — cultural, organizational, structural."
Writer surveyed 2,400 knowledge workers split evenly between C-suite executives and employees, all of whom were actively using generative AI tools at work. Last year's defining theme was tension — messy implementations, murky ownership, IT vs. C-suite turf wars. This year's theme is deeper: structural dysfunction. 73% of CEOs report their AI strategy is causing them stress or anxiety, with 38% describing it as high or crippling. 61% of executives fear losing their job if they fail to lead their organization through the AI transition. And when you look at why — 39% don't have a formal strategy to drive revenue from AI, and 75% say their company's AI strategy is more for show than actual internal guidance. 56% say AI has already created power struggles inside their organization, up from 42% last year. On the employee side: 29% of employees (including 44% of Gen Z) admit to actively sabotaging their company's AI strategy. 35% have entered proprietary or confidential information into a public AI tool. Two-thirds of executives believe their company has already suffered a data breach from an employee using an unapproved tool. The trust gap is the most damning number: just 35% of employees say their manager is an AI champion — and 75% say they trust AI more than their manager for certain work tasks.

WalkMe: The Tools Aren't the Problem
93% of AI spend goes to infrastructure and models. 7% goes to the humans using them.
SAP's WalkMe subsidiary surveyed 3,750 executives and employees across 14 countries. One third of employees hadn't used AI at all. Another 54% had bypassed their company's AI tools at some point to do work manually instead. The executive/employee trust gap is a 52-point chasm: 61% of executives trust AI for complex, business-critical decisions, versus just 9% of workers. On tooling, 88% of executives say their employees have adequate AI tools. Only 21% of workers agree — a 67-point gap. This reflects a structural misallocation that shows up across virtually every enterprise AI study: roughly 93% of all AI spending goes to infrastructure, models, compute, and tools. Just 7% goes to the humans who are supposed to use those things. That is a recipe for exactly the chaos the data is describing.

The Bottom Line
"There is a leadership crisis when it comes to AI. The companies that don't solve it are going to fail."
The through-line across all of these studies is that picking the tools and getting access to the models is not enough. The companies seeing results are designing systems and structures that support AI use — and the people using it. Two tiers are forming in the workplace: those who have gone all in on Claude Code or OpenClaw or Cowork or Codex genuinely feel like they have superpowers. Everyone else feels adrift and at risk of obsolescence. 92% of the C-suite say they're cultivating a new class of AI elite employees, and 60% plan to lay off employees who can't or won't use AI. AI super-users are roughly three times more likely to have gotten both a promotion and a pay raise in 2025 compared to non-users. The excited anxiety of enterprise AI is real — and it's a leadership problem, not a technology problem.