Startups are hiring far fewer people than they used to. In January of 2022, companies on Carta hired 73,000 people. In January 2023, they hired 40,000. In January 2024, they hired 32,000. And in January 2025, I think it's going to be like 20,000 or so. I am not one of those people that says AI is not going to take jobs. I think it kind of already is.

How are changes at venture capital impacting startups and scaleups? And what does this mean for software engineers? Peter Walker is the head of insights at Carter and Carter is used by more than 50% of all US startups to record funding and equity related data. Peter collected over a dozen interesting data points and charts about what is changing with tech startups.

Today we talk about how the number of startups getting funding has been steadily dropping even though the amount of VC invested has remained constant. Data showing the impact of AI and how it might result in smaller teams and more solo founders. what to look for before joining a VC funded company and what can help you thrive at a startup.

Now let's start with the venture capital ecosystem, how healthy it is and and how can we think of of health. In the US that is a venture is a subset of private equity. It's the standard things that people are familiar with of all these wonderful investors going around finding diamonds in the rough young companies that need capital.

Hopefully those companies then shoot up in valuation and they become the public companies that we love today. So you know all of I think all of the MAG 7 at one point or another took venture capital. The Mac 7 is a seven guess tech companies right? Is it Microsoft, Apple, Google, Nvidia? I think Mango is the new one where you talk that includes Nvidia as it the the acronym always keeps changing.

Yeah. It used to be fang even though it's now meta not Facebook but yeah. Exactly. Yeah, they're they're screwing with the names a little bit, so we got to keep changing the acronym. But those big uh seven tech companies are good examples of what venture capital is trying to do. They're trying to find really really young companies, give them a slug of capital, and then watch their valuations expand dramatically over time.

So VC, for a while, it kind of felt like venture capital was the quote unquote default way to build a startup. That's never actually been true. There's always been many, many more companies that don't take VC versus the ones that do. It's just a lot easier to talk about the ones that do take VC because as you raise money, there are sort of checkpoints along the way. So, you get seed rounds and series A rounds and series B rounds. Whereas, if you don't take any VC, nobody really knows where you are as a business. And so, it's a lot easier to talk about the venturebacked businesses than it is about the non-venturebacked ones.

And as someone who has been embedded in in the VC world, what do you think the main differences are? just thinking about as a software engineer or as an employee thinking about joining a VC funded company versus a nonVC funded one what might be something I know it's hard to generalize but but still let's let's try right like what what might be differences in terms of culture pace compensation those kind of things.

I'd say it actually boils down to one key difference which is venture is in the business of funding growth and not just regular growth but expansive growth so any sort of VC backed business will by definition be pushed by their investors and should have the outlook that we are going to grow super fast.

So for instance, the recent examples of the AI companies like cursor and others who are just ballooning up through different ARR metrics faster than basically any companies we've ever seen, those are perfect candidates for venture capital. If instead you have a business that's growing 20% a year, they're making money, it's a great business, but that isn't a candidate for VC because it doesn't project to have the growth rates necessary to make these bets worthwhile for the investors.

And that obviously has downstream impacts to compensation that has downstream impacts to what the company is focused on. You know, the pace of the actual work, the expectations both from the board and from the founders. There's a lot of changes that happen when you're beholden to investor expectations versus you are just building for yourselves and for your customers.

I worked at Uber at its craziest growth time and I worked at companies that might have had either venture funding or or or some of them just didn't like I think one of the things that I found is at at least at a very high growth startup like a VCR that is actually growing it felt kind of very exciting slashstressful and also a bit irrational and by this like at Uber like I remember we had an outage uh in India some of our our payments went down and we did a postmortem of like what what was the damage right like how much money did we lose and the engineer said like actually we saved the business a bunch of money I'm like what do you mean he's like well on every ride we lose an average of $2 in India because we're in growth phase therefore you know we miss 50,000 rides we save like $100,000 and I was like hold on like well actually that that's kind of true true and and this thing would have never happened at another type of company and and you know like I I I I think you're right like it all goes back to to growth and and also I feel it's a mindset because sometimes growth does mean irrational things. For example, we like grew the business knowing we're losing a lot of money like it felt painfully much but in the end I look at where Uber ended up and yeah it it worked out for them for for that specific category.

Yeah, it's it's a great point that there's this inherent trade-off often times although AI is maybe starting to challenge this uh between profitability and growth. So if you are pouring every single dollar into growth, often times you're doing so in on a unit economics basis unprofitable ways. Whereas a normal business uh if you think of you know a normal retailer or a restaurant or whatever they would never consider, oh we need to grow so fast that we are actually losing money on every time someone comes in and orders dinner. That doesn't make any sense.

But the VC model is about using capital at the beginning, funding incredibly high growth rates and then reaping the rewards of those growth rates because the company has gotten so large that they dictate what's going on in the market. So I think actually Uber is perhaps the example of when venture capital works well because now Uber is a very profitable company growing still growing very very quickly.

There are a lot of examples and this is where people kind of get mad at VCs where a company will take venture capital and then it'll turn out that there just wasn't the growth rate there. It wasn't possible to grow as fast as the VCs needed. So the investors go eh whatever we're going to go to our next bet and the founders and employees are left to say well we have to build a company out of this thing that is no longer you know the favored child of these investors. And so there's sometimes are those sort of misaligned incentives between investor and founder.

So I think when we talk about the health of VC backed startups a lot of time we by health we kind of default to how much money is being raised by these startups. And if you just go on that basis, VC looks very robust and healthy. But those are typically very very power law outcomes. Meaning there's a tiny cohort of companies that actually end up raising a ton of the money. And right now that's you know we all can list them open AAI, XAI, etc. who are raising billions and billions and billions of dollars.

If you instead look at just the number of rounds raised instead of the amount of capital in each of those rounds, uh the picture looks a little bit less healthy. You know, early stage startups, so seed and series A startups, we saw an average of 7.4 rounds per day on Carta so far in 2025. That is about half of where we were thinking at in 2021, and it's gone down every single year since 2021.

So, so this means just fewer companies are are raising. It might be bigger rounds or or whatever, but like you know that that that sounds bad, right? Exactly. Fewer companies are raising. Now, why are fewer companies raising? There's a ton of reasons behind this. I think two fundamental ones. First, um, everyone, and by everyone, I mean all the VCs are looking around and going, "Wow, the world has changed. I used to think that a company growing 100% a year was a great company. Now it maybe I'm looking for 200% growth a year or 300% growth per year when I used to look for 100%. Because they have these examples of cursor and other places where these companies are blowing past the old benchmarks.

The second reason and maybe this is a good thing is there are a cohort of companies who are looking around and saying maybe we don't need to raise venture capital. Maybe we are going to build this business and try to get to profitability and then have a lot of options. Maybe we take outside capital, maybe we continue to bootstrap, whatever it is, but we are not going to make VC the default way we build. And that I think has definitely especially in Silicon Valley that did not used to be a very common thing and now it's getting more and more talked about in the ecosystem.

If you're talking about how AI has impacted startups, I think this is the single biggest trend from our data. And the trend is this. Startups are hiring far fewer people than they used to.

So if we just look at this uh showing a chart here with uh hires and departures, but if you just focus on the black line that is hires, in January of 2022, companies on Carta hired 73,000 people in a single month. That's a very high number. In January 2023, they hired 40,000. In January 2024, they hired 32,000. And in January 2025, I think it's going to be like 20,000 or so. And and just to confirm, how how do you track the hires? It's equity allocations. So if a new hire is granted equity, we do now connect into many many many his systems as well.

if you look from 22 to 23, that massive decline in hiring, is that AI? Probably not. That's mostly just people having less funding. Less capital means fewer hires. Easy. 23 to 24. Also true that funding is is a little bit harder. So that's definitely a capital question. Maybe a little bit of AI. But 24 into 25 and 25 into 26, we are just flooded with stories of companies that say, "Look, we have an engineering team of 10, instead of moving that to an engineering team of 11, each one of those 10 engineers is just far more productive than they used to be because of AI tools. So we don't need to hire." And so that is sort of expanded across many, many companies over time, I think, is starting to be an explanation that makes a lot of sense to me. I am not one of those people that says AI is not going to take jobs. I think it kind of already is.

if you just look at the data as it is today, uh the pattern is is again very clear. So if we just focus on this series A column in 2022 on the day that you raised your series A round companies on Carter were about had about 20 to 22 full-time employees. Today it's more like 15 full-time employees and by the end of the year I think that's probably going to be more like 13 or 12.

So this idea that small teams are all the rage and people are trying to keep headcount low and grow as fast as possible just with the people that are currently in the business. It's I think it's very true and a lot of it can be wrapped up in this single metric that everyone across Silicon Valley is now talking about which is ARR per FTE. How much revenue do you have for each employee that works at the business as a measure of how capital efficient you are? That is a metric that many many many more VCs are asking startups for earlier and earlier in their life cycle.

It's not new necessarily in that it's always been a metric that people care about. What's different is the emphasis. So well if we go back to 2021 literally the point was grow as fast as possible and if that means hire a bunch of people hire a bunch of people. We don't actually care that much about the metrics were like monthly average users, total number of users, that kind of stuff, right? It wasn't just about revenue or just growth about ARR change. Yeah, totally.

And then obviously as funding declines, uh, businesses and their investors get a lot more concerned with how much money are you burning? What's your cash burn per month? Um, and so you're trying to remove that. And now we're in this place where, again, I think this goes back to the examples across the industry. If it's possible to build a company that has a hund00 million of ARR with 20 people, that's a more profitable, more capital efficient company than the same company with 100 million ARR that has 200 people.

So, this is data from Silicon Valley Bank. So, this isn't from Carta, but they're Silicon Valley Bank still an ongoing concern, I promise, in Silicon Valley. And this is data on how much ARR, so annual recurring revenue, does a startup have at series A. And in 2021, the median startup had about a million million and a half dollars in ARR. And in 2024, it's nearly $3 million. So it went from 1.3 or so to three and the high end the upper quartortile you know the companies that are doing very well exploded even further. I mean the companies right now on the 75th percentile for ARR have about $7 million of ARR at series A which is like six times as much in two than it was in two years.

It's wildly high. And so that's what investors are looking at. They're going, "What are going to be the generational companies of the future?" Well, if there are companies that are growing at this pace, you know, again, we get back to that original point. Okay, growth just doesn't cut it anymore. The metrics have gotten a lot higher.

How important are valuations? And by the way, what does price seed mean? In this chart, we have price seed, series A, series B, series C. I know what seed round is. It's the first typically the first time when you invest, you might have an idea or might you might have a product already. What is the price seed? So in this case we said priced seed round in to distinguish it between a seed round that is done on priced equity which is the normal you know you get down you sit in a room with an investor you come to an agreement on valuation which means every one of the shares of your company has a price you have a price per share.

另一种方式是通过 SAFE（Simple Agreement for Future Equity: 未来股权简单协议）进行融资，这种方式越来越普遍。SAFE 是由 Y Combinator（美国著名创业孵化器）推广的一种工具。它非常简单：投资者现在给创始人一笔钱，创始人承诺在未来某个时间点给予投资者公司股权。

这听起来很奇怪，为什么投资者会接受这样的交易？因为要为一个非常年轻、微小的初创公司确定一个准确的美元估值几乎是不可能的。SAFE 的好处在于，创始人今天就能拿到钱去发展业务，而投资者则获得了在公司发展顺利时获得股权的承诺。

估值泡沫：AI 与非 AI 公司的冰火两重天

高估值对创始人来说很有诱惑力，因为说“我是一家价值 5000 万美元公司的创始人”感觉很棒。在市场狂热时期，由于对公司未来的极度乐观，估值会飙升。

目前在美国，种子阶段公司的投前估值（Pre-money Valuation: 投资发生前对公司的估值）中位数约为 1600 万美元。投后估值（Post-money Valuation: 投前估值加上本轮融资金额）就是这个数字加上你筹集的资金。例如，如果你以 1600 万美元的投前估值筹集了 300 万美元，那么你的公司投后估值就是 1900 万美元。

这其实相当昂贵，甚至比 2021 年市场最火热时还要贵。2021 年是零利率、疫情红利等多种因素共同作用的结果。然后我们经历了 2022 年和 2023 年的低迷。但在低迷之中，ChatGPT 发布了，引发了 AI 的热潮。

所以现在的情况是，AI 公司正乘着新一轮炒作浪潮，前景备受期待；而非 AI 公司则陷入了低迷。因此，根据你所谈论的公司类型，风险投资市场要么前所未有地火热，要么感觉异常寒冷。

警示信号：过桥融资与估值下降

过桥融资（Bridge Round: 在两轮正式融资之间，为帮助公司“渡过难关”而进行的小额融资）通常由现有投资者提供，当公司未能按预期进入下一轮融资（如 A 轮）时，他们会请求更多资金以维持运营。

不幸的是，我们的数据显示，过桥融资通常不是好的赌注。在种子轮和 A 轮之间需要过桥融资的公司，最终成功进入 A 轮的比例远低于那些不需要过桥融资的公司。这很直观，如果一切顺利，你根本不需要请求更多资金。

数据显示，2020 年，约 33% 的定价过桥融资成功了。但到 2021 年，这一比例降至 16%，2022 年更是骤降至 8%。这意味着，如果你所在的公司正在进行过桥融资，根据最新数据，它成功进入下一轮的几率大约只有 8%。

另一个警示信号是估值下降轮（Down Round），即公司以低于上一轮的估值进行融资。从文化上讲，这通常被视为一种“承认”事情进展不顺的信号。由于初创公司的世界里，形象和制造兴奋感非常重要，因此估值下降对创始人来说可能极具挑战性。他们必须回去告诉员工，公司的价值比想象中要低，这可能会导致员工开始考虑离职。

AI 时代的新趋势：单人创始人崛起

AI 已经彻底改变了初创公司的构建方式。我们最清晰的发现之一是，单人创始人（Solo Founder）公司变得越来越普遍。随着创业成本的降低，以及单人能完成的事情越来越多，过去需要联合创始人的工作，现在可能一个人就能启动。

2024 年，在 Carta 上成立的初创公司中，超过三分之一是单人创始人公司，这是有史以来的最高比例。

然而，有趣的是，风投们仍然对投资单人创始人感到犹豫。虽然约 35% 的公司是单人创办的，但在获得风投融资的公司中，只有约 17% 是单人创始人。风投不青睐单人创始人的原因有很多：传统的“技术+商业”双核团队模式、关键人物风险，以及一个隐藏的观点——如果你连一个联合创始人都说服不了，你可能也无法吸引其他人才。

融资周期变长与“僵尸”初创公司

过去，初创公司融资的普遍法则是每 18 到 24 个月进行一轮新融资。但现在，这个周期已经大大延长。数据显示，从种子轮到 A 轮的中位数时间可能需要两年半，从 A 轮到 B 轮可能接近三年。这对创始人的明确启示是：你必须让手头的资金维持更长时间。

此外，从种子轮“毕业”到 A 轮的成功率也在变化。在 2021 年的繁荣时期，这个比例可能高达 50%。而现在，你应该预期大约只有 25% 到 30% 的种子轮公司最终能筹集到 A 轮。这意味着超过一半的公司将无法成功晋级。

一个棘手的话题是，有很多在 2021 年或更早筹集了种子轮的初创公司，它们仍然存活着，但基本上停滞不前。问题是，这些人是否应该关闭公司？硅谷文化中充满了“永不放弃”的故事，比如 Figma 和 Slack，但每一个成功案例背后，都有数百个失败的公司。

对创始人来说，如果业务确实无法达到风投所期望的增长，关闭它并尝试下一个想法，可能是一个更好的选择。投资者通常也愿意支持那些从失败中吸取了教训的创始人。

对员工和求职者的建议

作为一名工程师，在加入一家风投支持的公司之前，你应该像投资者一样思考。这家公司的独特技术优势或护城河是什么？在 AI 能够快速复制功能的时代，为什么这家公司能赢？

如果你加入的是一家种子轮或 A 轮的公司，你很大程度上是在押注创始人。你需要对创始人有信心，相信他们是你想追随的人。

在初创公司，除了技术能力，个人技能也至关重要。你很可能需要扮演“球员兼教练”的角色，亲自动手写代码。你还需要成为一个“瑞士军刀”，参与到纯代码之外的事务中，比如与客户交谈、评估市场。这正是初创公司的魔力所在：你不仅能提升编码技能，更能极大地拓展对整个行业和商业运作的认知。

成为顾问：期望与现实

成为初创公司顾问通常会获得少量股权。很多人期望能拿到公司 1% 的股权，但这非常高。根据我们的数据，2024 年 5000 名顾问的股权授予中位数是 0.25%。

作为对比，如今科技初创公司的第一位员工（通常是工程师）的中位数股权是 1.5%。考虑到顾问并非全职工作，每月只贡献几个小时，0.25% 的股权就显得合理多了。

对初创公司最有价值的顾问有两类：一类是技术顾问，拥有解决特定领域高难度技术问题的世界级专业知识；另一类是商业顾问，能够为公司引荐关键客户，直接带来收入。那些每月只开一次会、泛泛而谈商业计划的顾问，其价值远没有他们想象的那么高。