Key Takeaways

  • Engineering compensation in Silicon Valley has returned to peak 2021 levels, driven by massive capital flowing into frontier AI labs.
  • Matteo Franceschetti treats candidates cross-shopping Eight Sleep against Anthropic as an immediate hiring red flag because the operational reality and risk profile are completely incompatible.
  • Eight Sleep generates multi-hundred-million-dollar revenues with only 160 employees by demanding high individual output rather than hiring for headcount.
  • Franceschetti claims Eight Sleep achieves 5x the average Silicon Valley output per person, resulting in revenue per employee that tops Apple.

The Hiring Trap of Frontier AI Salaries

Silicon Valley compensation is spiraling again. Big AI labs with billions in venture backing are handing out packages that distort the entire hiring market. Franceschetti sees the pattern clearly: “We are back to 2021. And that I think is becoming a problem in the US, meaning salaries are going out of control.”

When a seed or growth-stage startup tries to match the cash and liquid equity of an OpenAI or Anthropic, it destroys its own burn rate. Early-stage startups cannot win a direct bidding war against companies backed by hyperscalers. Trying to compete on raw compensation attracts candidates who optimize for safety and market peak compensation over equity upside and ownership.

The Anthropic Red Flag

Franceschetti has a simple filter for candidates who compare Eight Sleep to frontier research labs: he does not want them.

“Sometimes people say, 'Oh, I'm at the final stage with Anthropic and Eight Sleep.' And to me, that is a really bad red flag. There is no way that these two jobs are in any way similar.”

An AI frontier lab operates as a well-funded research institution racing toward raw compute scale and foundational models. An operating business like Eight Sleep is building hardware, optimizing supply chains, running growth experiments, and selling hardware and software subscriptions. A candidate who views those two roles as interchangeable does not understand startup stages, role scopes, or personal risk tolerance.

“At Eight Sleep you should come to try to do 100x or 1,000x,” Franceschetti says. “The team size, there are so many details that are totally different. We actually want people that are very driven and ambitious.”

If an applicant treats a 160-person scaling company as a backup plan to an AI research lab, their goals do not match the pressure of building an operational business.

Lean Teams and 5x Output

Instead of raising headcount to match revenue growth, Eight Sleep caps its team size. The company reached multi-hundred-million-dollar revenue with just 160 people. Franceschetti attributes this efficiency to strict performance standards and small, high-density teams.

"Our revenue per employee is way higher than Apple. Our output is 5x what is average in Silicon Valley."

Adding headcount creates management drag, slows decision-making, and dilutes ownership. When you keep the total headcount at 160, every hire must produce outsized results. Startups that attempt to solve execution bottlenecks by hiring more mid-level engineers end up slowing down, while ultra-lean teams ship faster by keeping coordination costs near zero.

What to Do With This

In your next final-round interview, ask the candidate to name the other two companies they are evaluating. If their shortlist pairs your early-stage operational startup with a frontier AI lab or Big Tech monopoly, drop them from the pipeline. Spend your equity on builders who want raw upside and operational ownership, not candidates hedging between a venture bet and a research salary.