Sierra vs ElevenLabs: Why Tool Calling Beats Voice Models
Harry Stebbings and Cliff Weitzman break down voice AI: why Sierra owns enterprise GTM while ElevenLabs and Speechify battle at the API layer.
How to Build Your Own Data Center & Why Every Startup Should Do It
With Harry Stebbings, Cliff Weitzman · Sunday, September 6, 2026
Cliff Weitzman, founder and CEO of Speechify, joins Harry Stebbings to discuss the financial and operational reality of buying tens of millions of dollars in Nvidia GPUs instead of renting compute. They dive into Speechify's strategic expansion from consumer text-to-speech into enterprise B2B voice agents, how AI is transforming engineering workflows and compensation, and how GPU computing is revolutionizing personalized biotechnology.
Key takeaways
- 18 customer support AI startups raised over $100 million in the past 18 months, creating a crowded application layer led by Sierra and Decagon. Read more →
- Software engineering at Speechify centers on prompt architecture, verification, and agent steering, with Claude Code serving as the primary tool ahead of Cursor and Codex. Read more →
- Cliff Weitzman ran 15 consecutive weeks of multi-omics testing (genome sequencing, proteomics, and RNA analysis) on his brother to track autoimmune neuroinflammation. Read more →
- Renting an NVIDIA H100 GPU on cloud hyperscalers costs between $35,000 and $50,000 per year, compared to an outright purchase price of roughly $30,000. Read more →
- Cliff Weitzman admits avoiding B2B voice APIs was the biggest strategic mistake in Speechify's history, allowing ElevenLabs to capture the enterprise market. Read more →
5 articles from this episode
Speechify's AI Engineering Playbook: Hiring for Slope over Syntax
Cliff Weitzman reveals why Speechify hires math talent over legacy coders, uses Claude Code, and measures production shipping over token leaderboards.
How Cliff Weitzman Uses GPU Clusters to Treat Rare Diseases
Cliff Weitzman explains how pocket sequencers, AlphaFold, and private GPU clusters enable founder-led research on rare orphan diseases.
Why Speechify Buys NVIDIA GPUs Instead of Renting Cloud Compute
Cliff Weitzman breaks down why renting NVIDIA H100 GPUs costs 1.5x their purchase price annually and how Speechify allocates compute across 3 tiers.
Cliff Weitzman: The B2B Mistake That Let ElevenLabs Win
Speechify CEO Cliff Weitzman explains why skipping B2B APIs was his biggest mistake and how he plans to beat ElevenLabs on cost.
More 20VC with Harry Stebbings episodes
- Who Wins in the Next Decade of VC | Margin and Price Matter Less in a World of AI | Sameer DholakiaOct 25 · 6 articles
- The Future of Inference Market & Model Routing: Commodity or Increasing in Value? | LangChain CEOOct 18 · 3 articles
- Valar Atomics Sues Day One Ventures | Meta's Back to the AI Frontier | NVIDIA's $30B Reflection DealOct 18 · 5 articles
- Atlassian CEO: Enterprises Won't Trust Chinese AI | Why The Jury’s Still Out on Specialised ModelsOct 18 · 6 articles