Hacker News Reader

Comment by mrkurt | Hacker News Reader

I don't want to sound flippant, because this is hard as fuck, but the profitability path for us is reasonably simple: have good unit margins, attract customers, help them grow.

We have good unit economics. The riskiest, most terrifying thing we've done is start with our own hardware.

For dev focused infrastructure, what we need to do is attract a lot of devs, get them to take us to work, and then help their employers build better stuff over a long period of time. This takes _forever_ because companies-with-procurement-processes are designed to not buy critical infrastructure from startups.

It's working for us so far, even if it's not for you yet.

View on HN

The part I don't understand is the combination of these statements you posted today:

> there's a market price for this stuff and we don't have pricing power.

> We couldn't raise prices for VMs even if we wanted to

> the profitability path for us is reasonably simple: have good unit margins

How do these square up? How can you reliably maintain good margins if your competitors set your prices? And your competitors have the scale to, for instance, design and manufacture their own hardware rather than buying off-the-shelf chips.

I would think the key would be not to accept being a commodity whose prices can be directly compared with other hosters.

I was hoping Kurt was going to do an extended spiel on hardware margins and bandwidth pricing. You should keep needling him for this, because it's super interesting.

I want my ip address blog post!

Please Kurt!

How hard is it to hire people who understand how a physical server even works these days? Probably most engineers under 40 have never touch a physical machine, under 30 maybe have never seen an ethernet cable...

I suppose you are being facetious because what you said sounds absurd.

I am generalizing for a few lulz. But I've also worked with people who have never seen a physical server - they've told me. And mac dropped the ethernet port from their laptops years ago. I've worked in offices where everything is on wifi. So I'm sure you could find quite a lot of younger devs who have literally never seen an ethernet cord. Fwiw I've never seen token ring hardware.

I'vve consulted for companies of all shapes and in many different industries and when it comes to container orchestration there are many hard requirements that platforms like Heroku or Render don't address. If you can carve out a niche for SMBs then great. Maintaining that lead in such a competitive segment with price sensitive customers is hard.

I have no doubt that fly.io is a market leader right now just judging just by optics, but I think it is hard to maintain that lead in this segment just as I think it may be difficult to go upmarket with this type of product.

What are the biggest unaddressed concerns?

If your customers are really SMBs it’d make sense to partner with low-code/no-code platforms.

Those things are so slow (for reasons which escape my small brain).

Probably because they build their margins by overcharging to run the resulting apps on oversubscribed AWS instances.

Its tough when revenues are costs are mismatched.

I'm genuinely curious, how does the hardware investment make sense for a SaaS platform when bare metal can be leased as a service from Equinix and such? It the difference in TCO between bare metal you own and bare metal you lease really that large after factoring in the cost of all the logistics and labor?

Reply on news.ycombinator.com