Breakout startups are basically startups that have found product-market-fit and are growing at an incredible pace. So think of the OpenAIs and Databricks of the world. Back in the day these would've been baby Google and baby Facebook. Here's a link for more examples:
https://rocketshipstartup.com/#the-2024-list
Sam Altman, Dustin Moskovitz and a few others strongly suggest people in any stage of their career to join these startups.
I revisited Taro's course on picking a company, and there doesn't seem to be much info on them. The course does talk about startups but I feel that startups have so many stages that it's hard to make generalizations across them.
From my understanding, it seems like breakout startups are a great place to join:
Pros
Cons:
The Taro course suggests new grads to pick big tech. Funnily enough Sam Altman laments that someone picked big tech over a breakout startup. I'm not saying that these two pieces of advice are contradictory, but what factors should you consider if you're deciding between big tech and breakout startups?
Well if you can get into OpenAI or Anthropic, absolutely. But if you're part of the 99.99% that can't get in, it's okay. I knew someone who got into Midjourney and he had to be so so proactive to even get noticed.
Taro can give you all the resources in the world, but the harsh reality is that there is very little outside information on private companies. So if I were in your shoes, I'd try for big tech simply because there is much less unnecessary risk.
You never know if your startup
Especially if you're looking for a quick cashout, founders will totally sell you that dream to get you to join.
Some harsh truths:
If you think this would never happen to a break-out startup, it can at any moment. Meta is absolutely killing every other open-source LLM startup, including Mistral (which had the world's best cost/performance LLM for a few months).
Now, reasons to work at a startup despite all the instability:
Ultimately, nothing from the startup will likely remain. The equity can be worth pennies, nobody will remember how many all-nighters you pulled, and people don't care about the startups that never made it. What remains is your skills and experiences.
Well, at least that's what I tell myself so I can sleep at night
You never know if your startup
- is about to hit a major down round
- doesn't have product-market fit
- has 2 weeks of payroll left
- even has a good engineering culture
If you think this would never happen to a break-out startup, it can at any moment.
I'm not saying it's impossible that breakout startups will implode, but the chances are much lower than a pre-pmf startup. The point is that the risk/reward ratio of breakout startups are much, much better than earlier-stage startups, and you even get strong career growth. The tradeoffs, as I mentioned, is that you do take on that risk of the startup going under, even though it's smaller than most startups. The other is wlb, which we both seem to agree on.
Meta is absolutely killing every other open-source LLM startup, including Mistral (which had the world's best cost/performance LLM for a few months).
I wouldn't consider Mistral a breakout startup. Maybe to ensure we're on the same page, let's stick to discussing startups from here:
I think most of those startups are definitely later stage, which are much safer. It's just that there is the constant risk of being no longer seen as that "breakout startup". For example, OpenAI releasing a killer search engine would hurt Perplexity.
Mistral this time last year would have been considered a breakout startup :/
I'd just be wary of any of the generative AI startups on that list, some of them have serious customer churn (so they wouldn't be considered breakout startups) and are having trouble raising a next funding round.
I feel like this question is bit off as companies like OpenAI and Databricks are clearly not really startups anymore. OpenAI in particular isn't as it's going to raise at a $150 billion valuation, bigger than 95% of large tech companies (literally 2x bigger than PayPal which is a storied, legendary, huge tech company). I had a coworker who described these companies well: These companies are "pre-Big Tech" as everyone knows they're going to reach that $10B+ scale someday with similar pay/prestige.
You would obviously go to these companies over FAANG - They will almost certainly pay more in the medium/long-term, and you'll have way more scope. I would recommend the same thing in the past if an engineer had a Google offer vs. a Series D Robinhood offer or something. Same with pre-IPO Uber, DoorDash, Pinterest, Airbnb, etc. Notion is another great example of a unicorn that is almost certainly going to succeed - Definitely take that over Amazon/Google/Meta if you have a Notion offer as a new grad.
Where it gets tricky is when you're outside the super, super top tier with hot Series A/B/C companies that haven't raised at a +$1 billion valuation yet. So they clearly have a lot of potential, but they're not as proven as something like OpenAI. At that point it really go either way, but I would recommend Big Tech. Even among Big Tech, there are layers: I personally consider Meta the best FAANG compay and it's a considerable amount better for early career growth compared to something like Microsoft.
A question I have is how big is the difference in the brand value between your average breakout startup and a big tech? Is it negligible?
This is a tricky question as a "breakout" implies non-average. So I looked through the first ~35 companies on your linked spreadsheet and used that as an average.
If we're using that as the measuring stick, I do think there's a big gap in Big Tech's favor. For example, I have never heard of ClickHouse (which is on that list), and if I haven't heard of it (it's literally my job to keep tabs on the overall tech scene and I'm a Silicon Valley native), there's a good chance many recruiters haven't either.
That spreadsheet is pretty wild as it has companies like Notion (probably 2x stronger brand value than FAANG) and then stuff like ClickHouse. Huge variance, haha.
Funnily enough Sam Altman laments that someone picked big tech over a breakout startup.
Sam Altman is one of the most biased and privileged people to say this LOL. Dude literally runs OpenAI. The second ChatGPT launched, the brand value of getting an OpenAI offer instantly jumped to 5x that of a traditional FAANG offer. OpenAI is the most exciting company to work for right now, and it's not even close. It's reflected in their compensation too as OpenAI pays 50%+ more than Meta and Meta is already known as a company that pays top-of-market even within FAANG. That combination is just so incredibly potent. Literally nobody in Silicon Valley thinks that a Google/Meta offer is better than an OpenAI one due to them being more stable/traditional.
Thanks for the insight! I agree with you that the list has huge variance. For example, I was surprised not to see Stripe or ScaleAI on the list. And, as you mentioned, there are some pretty hidden startups like Clickhouse. There are also startups in between. For example, I've only heard of Applied Intuition once or twice and it's like top 10 on the list.
I like how your friend describes it - "pre-big tech". Can you talk a little bit more about what makes a startup pre-big tech? I can definitely see size being a huge factor, but that doesn't show the full picture (for example, Anthropic has 800 people and everyone knows them, whereas Remote has 5K people and I've never heard of them).
Pre-Big Tech constraints:
Number of people will vary a lot among these companies. Some companies are more aggressive than others. For example, Robinhood was pretty conservative with hiring up until their Series D. Other companies will start trying to 3x - 5x headcount year over year after raising a large Series A. This is why some companies with $5B valuation will only have like 100 people while others will already have 500+ people. It's less about the current state and more about the momentum and future trajectory.
Hey Alex, not quite related to this post but I remember you mentioned there is a Leetcode course in the making. Any approximate timeline when we can expect it? Thanks!
It's half-baked (the descriptions are placeholders or empty), but you can watch it here: [Course] Master The Data Structures And Algorithms Interview