A16z partner: High ad spend is a warning sign for AI startups

This article summarizes an episode of EO’s video series featuring Anish Acharya, partner at Andreessen Horowitz.

Anish Acharya, a partner at Andreessen Horowitz / Photo credit: Andreessen Horowitz

High marketing costs in AI are a warning sign that a software product is weak. Anish Acharya, a partner at Andreessen Horowitz, warns startup leaders against buying growth to hide low demand. He argues that true value comes from building useful tools and charging premium prices.

AI features must be transformative, not incremental

Moving beyond paid acquisition requires developers to build tools that alter how people work. Acharya notes that advertising masks software flaws, adding that “10, 50, or 100 lead bullets never equal a silver bullet.”

Teams must evaluate their development pipelines using these criteria:

• Target leaps: Products must be 100x better than existing options to survive.
• Cut average ideas: Releasing updates will not convince users to pay premium prices.
• Focus engineering talent: Direct developers to solve problems before hiring salespeople.

Narrow specialization forms a new competitive moat

Since general-purpose tools invite competition from tech giants, smaller companies must adopt a targeted development strategy:

• Niche targeting: Solve specific problems for a tiny customer base.
• Workflow automation: Eliminate manual labor to justify premium pricing.
• Defensive specialization: Focus on industry details to prevent competitors from copying the software.

This focused approach generates revenue without needing a global audience. Acharya says that “charging 41,000 people US$200 a month is a US$100 million [annual] run-rate business.”

Premium pricing validates market demand

Monetizing those audiences requires abandoning market estimates in favor of price testing. Founders must demand upfront payment because offering software for free often masks a lack of demand.

Establishing premium rates grounds a business strategy in economics by enforcing validation steps:

• Design a package that guarantees a valuable result for the buyer.
• Monitor usage over several months to ensure customers stay subscribed.
• Track if users request features to solve complex problems.

Acharya pushes teams to test this by asking founders, “What is the US$1,000-a-month SKU of our product?”

⚖️ The other side: