Enterprises will be Slow to Adopt AI

Welcome to the hundreds of new subscribers who have joined over the past few weeks. In today’s free post, I take a look at enterprise adoption of AI.

And if you like what you read here and you’re not yet subscribed, consider subscribing. Most of my posts are free though deeper dives tend to be paid.

---

I’ve previously written about enterprise adoption of AI, asserting that it will be a trench war, not a blitzkrieg. Some conversations I had over the weekend reaffirmed that view. What follows is a short, high-level overview of why I think enterprise adoption of AI will occur over a longer timeline than many people in Silicon Valley expect.

There is the notion of the fully automated firm, in which AI(s) pervades all workflows across a company, and manages the company’s operations devoid of human intervention. While this type of thing is possible in theory, in practice, we are not likely to see a fully automated firm any time in the next decade. At least not for a firm of any notable size or complexity.

Silicon Valley’s vision and enterprise reality around AI adoption diverge not just temporally, but in terms of deep institutional, incentive, and epistemological misalignments. Let’s break it down structurally.

I. The Silicon Valley View: Accelerationist and Theological

Core Assumptions:

• AGI is inevitable, perhaps imminent.

• Once AGI arrives, adoption becomes a foregone conclusion.

• Enterprise behavior will rapidly change in response to software that can think.

• Startups that integrate AI into workflows will disrupt incumbents by default.

Motivations:

• Narrative fits the venture model: high risk, high growth, short-term return.

• Founders are incentivized to exaggerate traction and minimize integration friction.

• VCs tend to overindex on technical possibility, not organizational path-dependency.

Theological Impulse:

There’s a subtle religious belief at work: AGI as salvation. Once it arrives, all shall be transformed. This removes the burden of grappling with how enterprise processes change. Just wait for the Messiah.

II. Enterprise View: Cautious, Path-Dependent, Risk-Averse

Reality on the Ground:

• Enterprises don’t adopt technology. They adopt process change, and that’s expensive.

• Procurement cycles are long. Liability, compliance, and explainability matter more than capability.

• Integration costs, not inference costs, are the bottleneck.

• Enterprises want robustness, not cleverness or hallucination.

Organizational Inertia:

• AI models are non-deteriministic, which violates enterprise expectations around auditability.

• “Automating workflows” means reworking entrenched human hierarchies, union agreements, and regulatory frameworks.

Enterprise CIO Mindset:

• “We don’t need magic. We need reliability, compliance, and clarity.”

• They’ve seen the hype cycles: big data, blockchain, IoT, and more. They’re skeptical and risk-averse.

III. Why Enteprises Understand Themselves Better

Enterprises are a better judge of their own adoption pace than outside optimists. This is because:

1. Enterprise friction is endogenous, not exogenous: it’s built into who they are.

2. Silicon Valley personalities can model capability fit, but not sociotechnical entrenchment.

3. Enterprises don’t optimize for speed. They optimize for minimizing downside risk.

4. The bigger the org, the more power middle management has to quietly sabotage tech rollouts that threaten them.

IV. Deeper Asymmetries Driving the Bifurcation

1. Epistemic Bubbles

   1. Silicon Valley people mainly talk to other founders, demo day pitchers, engineers, and venture capitalists.

   2. Enterprise buyers mostly talk to consultants, compliance teams, and regulators.

2. Temporal Mismatch

   1. VCs and founders operate on 5-10 year cycles.

   2. Enterprises deploy software on 15-20 year timelines.

3. Incentive Misalignment

   1. Silicon Valley wins when adoption is hyped early.

   2. Enterprises win when adoption is quiet, boring, stable, and safe.

V. The AGI Fallacy

Many Silicon Valley people believe AGI will force adoption.

That’s a theological belief masquerading as strategy.

Reality:

• The more general a system is, the less controllable it is.

• Enterprises, especially in regulated industries, don’t want generality. They want modular, bounded automation.

• If AGI is real, they won’t be the first adopters. They’ll wait for it to be boxed, validated, and regulated.

VI. Consequences for Builders and Investors

Implication 1:

• AI startups targeting enterprise must focus on integration, not capability.

• The startup moat isn’t the model. Rather, the moat is change management strategy.

Implication 2:

• The AI enterprise playbook looks more like Accenture + fine-tuned LLMs than OpenAI-as-SaaS.

Implication 3:

• Real value accrues to those who solve for governance, auditability, latency, and human-in-the-loop workflows, not those chasing SOTA benchmarks.

The Valley is predicting what they want to be true, from within an echo chamber of incentive-mirroring peers. The enterprises are simply living inside the constraints that govern them: liability, regulation, legacy code, and political risk.

They are slow for a reason. And that reason won’t go away just because GPT-5 can pass a bar exam.

Coda

If you enjoy this newsletter, consider sharing it with a colleague.

Most posts are public. Some are paywalled.

I’m always happy to receive comments, questions, and pushback. If you want to connect with me directly, you can:

• follow me on Twitter,

• connect with me on LinkedIn, or

• send an email to dave [at] davefriedman dot co. (Not .com!)