The AI Era Was Never Easy, and Investors Are Raising Expectations

By Maor Farid 

Back in 2021, simply mentioning artificial intelligence on a pitch deck was often enough to secure funding. Small and mid-sized businesses rushed to adopt these technologies, purchasing them rapidly. I observed companies successfully onboarding clients and raising capital, only to watch user engagement plummet soon after. Retention rates dropped sharply.

A significant portion of those offerings were superficial: they mirrored existing traditional software while adding an unnecessary veneer of AI. Others saved a minor amount of time on functions that were entirely inconsequential to daily operations. Achieving a 20% gain on a trivial task generates an impressive demonstration, yet results in a canceled subscription.

True, effortless artificial intelligence triumphs never really existed, even during periods when capital flowed freely. The market is currently correcting for that reality. Enterprises delivering genuine value are expanding at a rate unmatched anywhere else in enterprise software. For investors, the central challenge is identifying which of these firms can sustain their customer base — and discovering whether upcoming initial public offerings will highlight the divide.

What makes an AI business defensible today

Before my co-founder and I developed any software, we spoke with over 900 mechanical engineers ranging from entry-level staff to vice presidents. We inquired about how they spent their working hours and which tasks they would gladly pay to eliminate. Those 900 discussions ultimately defined my approach to constructing an AI enterprise.

First: the performance enhancement must justify the financial commitment. Saving 10% on an infrequent chore rarely transforms operational workflows. Conversely, compressing a critical business process from weeks down to minutes, or generating substantial financial savings, certainly does.

Second: industry-specific knowledge. Frontier models are rapidly evolving into standard infrastructure. Almost anyone with an API key can tap into a high-performing large language model. Commercial success depends on mastering a specific sector thoroughly enough to resolve challenges that generic applications cannot touch.

Third: proprietary context. Consider physical engineering as an example: every organization that manufactures tangible goods has amassed decades of technical expertise. Much of this wisdom remains trapped inside legacy blueprints or locked within the memories of veteran staff. Foundation models lack this specific intelligence in their training data. An AI system unable to access this data will struggle to become indispensable to the enterprise. This competitive moat widens as the software becomes tightly woven into client workflows, making substitution increasingly expensive.

What this means for funding and IPOs

Upcoming artificial intelligence initial public offerings will subject these underlying business models to much harsher scrutiny. Public market investors will analyze net revenue retention alongside gross margins to determine whether rapid top-line growth converts into a viable commercial enterprise. Furthermore, they will check if support costs diminish as the company scales. The benchmarks they set will shape the expectations private investors hold for early-stage ventures.

I already notice this shift within my own fundraising dialogues. Several years ago, hitting $1 million in annual recurring revenue was a major milestone. Today, a newly launched product can reach that $1 million mark within roughly twelve months, only to fail a year later. Consequently, annual recurring revenue by itself no longer provides investors with sufficient insight.

Investors now focus heavily on expansion metrics. They want to ascertain whether clients increase spending following initial implementation. Expansion serves as our most reliable indicator that an application has fundamentally altered corporate operations: clients have witnessed its tangible value and allocated a greater portion of their own budgets toward it.

Over the next one to two years, I anticipate capital will continue directing toward organizations that merge deep sector expertise with dramatic improvements in core business functions, alongside access to proprietary client information. Such companies might scale through hands-on deployments rather than viral self-service adoption, yet they retain the capacity to expand within accounts that place trust in them.

Firms whose offerings amount to thin wrappers over third-party models — backed by flashy client rosters and retention metrics they prefer to keep hidden — will discover that securing their next funding round is significantly harder than the previous one. I would personally choose steady, healthy growth over five years rather than spectacular expansion lasting merely five quarters.

---

Maor Farid is the founder and CEO of Leo AI, the first AI for mechanical engineering — a large mechanical model for physical product design. He conducted AI and mechanical engineering research at MIT as a Fulbright postdoctoral fellow and became the youngest Ph.D. graduate in the history of the Technion – Israel Institute of Technology. Farid has built a community of more than 60,000 engineers and supports underserved youth through his nonprofit initiative.

Related Crunchbase query:

• Global Venture Funding To AI Startups In 2026

Illustration: Dom Guzman

Frequently Asked Questions

Why did many early AI products struggle with customer retention?

Many early products were generic solutions that merely duplicated traditional software with superficial AI features, or they saved minor amounts of time on tasks that were not critical to daily operations.

What makes an AI business defensible today?

A defensible AI business requires enhancements that justify the investment, deep domain expertise to solve niche industry problems, and access to proprietary context that foundation models lack.

Why is ARR (Annual Recurring Revenue) no longer enough for investors?

Because new AI products can hit $1 million in ARR quickly and shut down just as fast, investors now focus on expansion metrics to see if customers are increasing their spending over time.

What do upcoming AI IPOs mean for private startups?

Public market scrutiny on net revenue retention, gross margins, and scaling support costs will establish new benchmarks and influence what private investors expect from earlier-stage companies.

Who is Maor Farid?

Maor Farid is the founder and CEO of Leo AI, a former Fulbright postdoctoral fellow at MIT, and the youngest Ph.D. graduate in the history of the Technion – Israel Institute of Technology.