1

Managing financial risk while working at a frontier AI model startup

Profile picture
Senior Software Engineer at Taro Community2 months ago

Context

Thanks to Taro, I managed to get an offer from a frontier AI model company. When I saw the $500K worth of annual ISO component in the offer, I thought I nailed it. Until I digged further into the details.

Problem

I'd have to shell out $500K each year to purchase my vested stock options. My base salary minus tax and expenses would be around $100K. So I won't even be able to buy all the vested stocks unless I sell off my less-risky, diversified investments. This means I'd have to live a frugal life and worry about making $500K by the end of the year to afford the vesting grants. All this feels like too much risk accumulated around a single company (which is never a good sign). The startup offers stock buy back every 6 months but I'd lose out on long-term capital gains taxes if I sell off quickly without holding for a year.

None of the frontier AI startups have a clear plan to justify the sky-high valuation but I see a lot of senior, smart people flocking to these companies (which gives me hope). The reputation of the funding agencies participating in my company's funding round further raises my hope and begs to ask the question "How can so many smart people and amazing funding agencies get it wrong all at the same time?". The more I think about it, the more I am losing sleep.

Question

I am not the first person to signup up for a high-risk high-reward startup. What strategies does the employees at these companies typically follow to manage their risk?

72
3

Discussion

(3 comments)
  • 1
    Profile picture
    AI/ML Eng @ Series C startup
    2 months ago

    Run.

  • 1
    Profile picture
    Thoughtful Tarodactyl
    Taro Community
    2 months ago

    Dont do it. Also if you are in cali or other high cost of living place, 100k will not be livable.

    but I see a lot of senior, smart people flocking to these companies

    Does not mean much. AI model is needs the tail end of folks in AI. See if there is someone at the level of Andrej Karpathy, if not the company has no value.

    Also with Meta making Llama open source, and xAI creating 100k GPU training cluster, no startup can compete with that. Llama will make charging for models really hard, and xAI might have the best model if the model scales with size of training cluster, as no one else can create a cluster of more than 32k.

    • 1
      Profile picture
      Senior Software Engineer [OP]
      Taro Community
      2 months ago

      It is xAI 😬