i'm an early-stage vc - the author's analysis on "number of funds" (specifically VC funds) is accurate. the overall volume of venture allocation has also slowed considerably if not decreased (which is totally expected in a higher interest rate environment).
2021-2022 was a total blip on the screen zero interest rate era thing.
i'm not seeing considerable slowing of new startup development, quite the opposite actually w/ AI. this is for a few reasons:
- accelerators are filling the gap; the accelerator model is actually quite efficient in the early-stage spectrum (it needs some further innovation). there are a huge number of AI accelerators and programs now; and further
- most of the capital going into VC is just being further concentrated into the large Multistage firms like A16Z, Accel, Sequoia, General Catalyst, etc... all of these firms are realizing they need to win deals as early as possible so have multiple seed programs: accelerators, incubations, scouts, fund-of-fund allocation, geographic funds, university focused sub funds, etc...
- overall great founders & startups are truly just exceptional so statistically there just won't ever be that many. venture will always be a cottage industry of sorts. in this form - "venture" equates with "growth"; there can only be 1 category leader by definition and venture is meant to capture this. 2021-2022 overall venture market was too big.
- AI is making startup creation many multiples more efficient. we saw this w/ the advent of the cloud, where startups used to need $2-3M "to buy servers" and 2-3 years to ship a product in 2010, by 2015-2020, they really only needed $3-500k to get a product to market. we're going to see that number come down considerably (unsure if it will be 30-50k, but definitely a lot lower).
- we're also seeing the new wave of the 10-person unicorn (billion $ company); these companies will raise a lot less cash, so will result in higher multiples on the original investment.
- i think the overall distribution of returns will look different on a portfolio basis in 2025-onwards. with power law, we expect to see super long-tail concentration on the 1-2 companies that yield 99% of the return to a portfolio, but i suspect we'll start to see some mitigation of that effect with more companies yielding positive outcomes. this might mean that there's less of a reliance on portfolio construction to generate risk-adjusted returns and that there could be more of a democratization of early-stage investing where we see 10-100x the number of startups and founders. that warrants a longer analysis, but as someone just bullish on startups and everyone being a founder that possibility is very exciting to me.
One consideration not mentioned is around developer sophistication. Steve alludes to the expansion effect of CodeGen ("there are millions and maybe billions who are jumping at the chance to code"), but doesn't consider that the vast majority of these people don't know about arrays, data structures, memory, containers, runtimes, etc, etc...
To me, that's the most important consideration here. Are you targeting professional devs who are enhancing their current workflows iteratively with these improvements? Or re-thinking from the ground up, obfuscating most of what we've learned to date?
Maybe we need to trudge through all of these weeds until software creation hits its final, elegant form where "Anyone Can Code".
Maybe the old Gusteau quote is actually fitting here:
"You must be imaginative, strong-hearted. You must try things that may not work, and you must not let anyone define your limits because of where you come from. Your only limit is your soul. What I say is true - anyone can ̶c̶o̶o̶k̶ code... but only the fearless can be great."
Love the movement and glad there are founders out there pushing the envelope for their team.
(aside: 51 points but only 1 comment? It's a front-page worthy article, but sort of feels like there's some vote gaming happening. I've never seen 50 points w/ 1 comment.)
Are annual plans really that difficult to implement? Doesn't seem like it should be that difficult unless you have complex pricing behaviors. What are some of the most common stumbling blocks with transferring a plan from monthly to annual?
Check out https://anycart.com/ - they scrape available delivery slots for you. Also can save you a lot of time / effort and money if you prefer to order by recipe although lots of ingredients are out of stock. (disclosure: investor)
Not surprising to see the common complaints from learn-to-code authors about disrupting their courseware. We heard this a lot at Nitrous.io and struggled with the importance of the beginner market to our business.
When considering Nitrous as a viable business, I'd talk a lot about the "developer sophistication spectrum" and the challenges of one single product or service attempting to meet the needs of a lot of different types of developers.
On the newbie side of the spectrum, serving the hot "learn to code" market means scaling your potential market size by orders of magnitude. There is some product-market fit here as newbies don't really have substitutes ("what's a development environment?" they'd often remark), but the SaaS economics of selling tooling to newbies was atrocious.
Selling to learn-to-code means you're dealing with an incredibly fickle audience where 95% abandon their plans to become a professional programmer within a few months. The other ~5% who become full-time programmers are dedicated enough to their craft to learn about their OS and their options to customize the local development workflow. So they naturally also churn.
(I don't have any knowledge of the market, but I'd imagine courseware providers attempt to charge 100% up-front to account for the extremely high churn. At least, that's how I'd charge.)
So basically all the cloud IDEs are getting hundreds of thousands of signups from a lot of newbies saying "We love [Nitrous, Cloud9, Koding, etc...]!" but not wanting to pay for the infrastructure and churning at unsustainable rates.
On the less sophisticated side of the spectrum, I think there is potential for a viable cloud IDE business, but I think it needs to be closely coupled with a content platform like Treehouse, Coursera or CodeAcademy. I haven't looked at any of them recently, but I wouldn't be surprised if they have in-house teams working to improve the editor experience and provide stateful experiences with dedicated cloud compute & storage. We had a tightly coupled integration with the Flatiron School and it was a pretty solid experience but just wasn't a big enough business for us to scale. So in reality these businesses really just look like a content / courseware business that has a really great cloud development experience. But it's clearly built for people learning to code and they're paying for the courseware, not for the editor.
As you move up the sophistication spectrum, developers begin to experience "cognitive dissonance" when considering how much their time is really worth. That is, when they know how to setup, configure or troubleshoot something themselves, they underestimate the time they spend every month performing those tasks. We spent a ton of time doing deep customer research with excellent engineering teams at Airbnb, LinkedIn, Shopify, etc... You'd be extremely surprised at just how much time it takes for the average developer at a top-tier engineering org -- in some cases, new developers took 3-4 weeks to setup their dev environment. But after setting up a new environment the other dev ops problems start to spider into a web of complex and proprietary issues that are difficult to create compelling marketing / sales presentations. It's like - everyone knows it sucks and it's broken, but nobody quite knows the solution. Which is why a lot of the solutions emerge from open source projects that solve specific issues organically and then expand into powerful platforms that cohesively solve a set of interesting ops problems (e.g. Hashicorp).
This is an oversimplification of the complexities of the developer market - as there is also a spectrum of sophistication within the professional developer market itself. The "intermediate" professional developer tends to be the best market fit right now for cloud development / IDEs, as they often are self-taught and know how to code, but are often not as versed in debugging low-level issues, but usually are more price sensitive to their more sophisticated counterparts (who don't want to use the service in the first place).
In any case, I remember reading a HN comment about the nitrous.io shutdown [1] and feeling bad about not opening up more so I suppose this will provide some color. People loved our service and we honestly loved building it, but business is hard and we weren't able to uncover the right strategic focus. Hopefully Coursera, Treehouse, CodeAcademy, etc... will continue to fill in the gaps for the beginner market - but since those will be tightly coupled with their courseware, it's going to be a difficult spot to be in for the independent educator who is attempting to monetize their own material.
It’s easy to forget that good ideas take a lot of time and iterations until they become magical. Great things do not happen overnight and without much sacrifice. The majority of people would give up the first time the community dismissed their prototype.
We're working on a static site hosting service called PubStorm (https://www.pubstorm.com) that allows SSL on custom domains. It's free; we're also working on some powerful features that will set it apart from gh-pages.
Entrepreneurship is more about conviction, determination, and will than ideas and market opportunities.
To me, this letter is basically saying you're just not up for the challenge. That's fine -- it's certainly better to admit it now, and it's likely a rational decision.
But the best companies usually aren't created by rational thinkers. The best entrepreneurs can take a horribly broken system and contort it in ways that nobody else even thought of, and then execute on a vision to make it the new reality.
You're not telling your investors "this market is too tough, let's not go here", you're basically just saying "This market really needs to be un-fucked, but I'm not the guy to do it".
I had the same trouble setting up discourse, so I setup a template on Nitrous using Docker that you can definitely use to get Discourse up and running in 30 seconds. (I just confirmed, I went from no environment to running discourse in less than 30 seconds). Just `cd code/discourse && ./start-app`.
This, IMO is where Docker shines. It shouldn't matter if it's setup with a microservice 12-factor architecture or if everything is setup in a monolithic VM-like container. I don't have the patience for ops -- I just want something that works. That's the point of having isolated, replicable containers.
In any case, I encourage you to try out the discourse container on Nitrous. I was actually surprised it happens to be the least popular container for us. I assumed because it's such a pain in the ass to get started, that it would be more popular =p
Having a development environment and editor in the browser is definitely the way to go for students learning to code. Updating a browser is significantly easier and cheaper than purchasing a new machine. Kids learning to code shouldn't have to worry about specs, software installation, and OS configuration.
Nitrous, Cloud9, Koding etc... all have free tiers. We're working on Nitrous and definitely will continue to support students as best we can. We recently launched a native chrome application, and honestly with our chrome application a $200 chromebook can be a pretty amazing development machine, even for professional developers.
I've been working on customized EC2 instances, DO droplets and Nitrous (https://pro.nitrous.io) since 2012 and haven't looked back. There is the issue of connectivity, but I'm unproductive without an internet connection so it has worked well for me.
I use tmux and then connect to the session from work, home, etc... and setup ssh config (http://nerderati.com/2011/03/17/simplify-your-life-with-an-s...) so I have shortcuts to all of my remote machines. App environments are built with docker, snapshotted, and pushed to dockerhub.
Investment at 200x MRR for such a low margin business is extremely overvalued IMO. The only case for that type of multiple is if instacart grew more than 100% YoY (which they probably did).
Also of note and let's not forget -- $100M 2014 revenues are the "groupon-esque" pass through revenues -- Whole Foods, Safeway, etc... are also booking these as revenue.
Remember when Groupon, on the eve an IPO, cut it's reported revenue in half? Yeah, only a $300,000,000 difference.
> "On Friday, Groupon said it would change what it books as revenue after discussions with the Securities and Exchange Commission. It will now only count as revenue its commission on sales, rather than the total value of an online coupon."
There's a free plan on Nitrous Pro, so you can signup to reserve your free workspace (we're rolling out free workspaces, if you pay it's available immediately). We wrote about a few ways you can migrate your data to Nitrous pro here:
There are a ton of additional benefits of being in the cloud. You can backup your environment and replicate it in minutes, you can collaborate with other developers across the world really easily in real-time, and you can work from any device with a modern web browser. A lot of clients really prefer using Nitrous environments for their service providers because they feel as though they have more control over the developers they're hiring and their IP -- clearly there are still risks since there are fairly easy ways to pull code down but "owning" the machine that your contractors are working on is a step in the right direction for compliance and audit purposes.
We’ve found the biggest benefits are really around the collaborative use cases — sharing environments easily and working together remotely. It’s saved us a ton of time when we’re troubleshooting issues. Yes, you can use tmux over SSH, but not everyone wants to setup and maintain their own development server. I think that interacting with your environment using Nitrous Pro is a really enjoyable experience.
That said, we’ve still got a lot of work to do, so we really do appreciate all the feedback and support from the HN crowd.
I laughed to myself after reading the first few paragraphs. When I saw the image of young Chinese students I immediately thought of the literary use of myopia, not the medical condition. Given the source of the article was “Nature”, I assumed it was an article about the Chinese exploitation of natural resources for the sake of short term profits and the destructive effect on the environment; hence, myopia. Is that irony?
The tacit identification in the post is an interesting strategy. Why not just explicitly identify the individual? Are there legal implications here or is he trying to manufacture more drama by getting the reader more involved?
2021-2022 was a total blip on the screen zero interest rate era thing.
i'm not seeing considerable slowing of new startup development, quite the opposite actually w/ AI. this is for a few reasons:
- accelerators are filling the gap; the accelerator model is actually quite efficient in the early-stage spectrum (it needs some further innovation). there are a huge number of AI accelerators and programs now; and further
- most of the capital going into VC is just being further concentrated into the large Multistage firms like A16Z, Accel, Sequoia, General Catalyst, etc... all of these firms are realizing they need to win deals as early as possible so have multiple seed programs: accelerators, incubations, scouts, fund-of-fund allocation, geographic funds, university focused sub funds, etc...
- overall great founders & startups are truly just exceptional so statistically there just won't ever be that many. venture will always be a cottage industry of sorts. in this form - "venture" equates with "growth"; there can only be 1 category leader by definition and venture is meant to capture this. 2021-2022 overall venture market was too big.
- AI is making startup creation many multiples more efficient. we saw this w/ the advent of the cloud, where startups used to need $2-3M "to buy servers" and 2-3 years to ship a product in 2010, by 2015-2020, they really only needed $3-500k to get a product to market. we're going to see that number come down considerably (unsure if it will be 30-50k, but definitely a lot lower).
- we're also seeing the new wave of the 10-person unicorn (billion $ company); these companies will raise a lot less cash, so will result in higher multiples on the original investment.
- i think the overall distribution of returns will look different on a portfolio basis in 2025-onwards. with power law, we expect to see super long-tail concentration on the 1-2 companies that yield 99% of the return to a portfolio, but i suspect we'll start to see some mitigation of that effect with more companies yielding positive outcomes. this might mean that there's less of a reliance on portfolio construction to generate risk-adjusted returns and that there could be more of a democratization of early-stage investing where we see 10-100x the number of startups and founders. that warrants a longer analysis, but as someone just bullish on startups and everyone being a founder that possibility is very exciting to me.