How Speechify justifies spending more on AI tokens than salaries

This article summarizes an episode of 20VC’s video series  featuring Cliff Weitzman, founder of Speechify AI.

Cliff Weitzman, founder of Speechify AI / Photo credit: Cliff Weitzman on X

Running a modern software company comes with staggering new expenses, and Cliff Weitzman believes traditional management will not survive the shift. The Speechify AI founder and CEO argues that companies must cut any project that does not drive paying customers in order to afford rising AI compute costs.

Weitzman pushes leaders to drop rigid brand campaigns in favor of testing endless AI-generated ads, to ban time-wasting status meetings entirely, and to accept that paying for machine intelligence will soon cost more than employee salaries.

Sales are the only truth in getting customers

This sales-first mindset turns customer acquisition into a conversion equation. Weitzman views marketing as a tracking game and kills any project that generates attention but fails to convert into paying users.

“Growth is just an arbitrage game,” Weitzman says. “You are competing with every single other person in the world who wants to get their product in front of users.”

Weitzman believes companies should optimize content and distribution instead of relying on “gut feelings” or strict brand consistency. To achieve this, Speechify tests nearly a thousand AI-generated ads daily to see what actually converts.

“Only do things that result in conversion,” Weitzman adds. “If you don’t get people who actually convert on your product, there’s no point in doing the work that you’re doing.”

OpenAI ads will change targeting and privacy risk

Hunting for exactly what users want naturally drives marketers toward new platforms. As the industry shifts toward AI chat tools that capture personal details, companies gain access to a lucrative but legally risky ad environment.

The personal details shared during chat sessions create unprecedented marketing opportunities. Weitzman notes the potential is “Massive. Because it knows so much about you… OpenAI knows everything about your history and what you’re interested in inside of your psyche.”

High prices are irrelevant with perfect tracking
Exploiting this targeted chat data comes at a premium, but those upfront ad costs are irrelevant if the return on investment is proven.

“It’s okay to pay a high CPM as long as people convert and as long as you have attribution,” he explains. “So they [OpenAI] just launched the SDK (software development kit) for tracking. That’s really important.”

Tokens will become a regular payroll cost for engineering

Beyond acquiring users, integrating AI internally turns compute credits into a standard operational expense. Finance teams must adapt as rising model usage shifts software budgets away from salaries and toward machine intelligence.

Weitzman predicts, “Next year I expect we’ll spend more in tokens than we’ll spend on salaries. It’s atypical right now, but I don’t think it’ll be atypical in the long term.”

Speed is built by killing meetings and rewarding finished work

Computing costs decide who survives the AI boom

AI power increases inequality and pushes users to speak up