I hope they don't bet too much on AI, but I do think they deserve to grow more as a cloud provider, they need to focus strongly on following up on people's complaints about their existing services. I have seen many threads of people with complaints about their S3 compatible storage being very limited in its design, and potentially wildly insecure. If they are serous enough to be asking for roughly a billion dollars, they should really consider fixing up their current offering either first or at the same time. It's only a matter of time before some giant scandal has customers running away.

I say this having used DO since... 2012?

▲

enlyth 3 hours ago | parent | next [-]

They occupy some weird middle ground where they've become too expensive for hobbyists but not good enough for enterprise use

	▲	giancarlostoro 3 hours ago \| parent [-]
		> but not good enough for enterprise use This goes back to my suggestion to strengthen their current offering instead of making a giant frankencloud. They're in an interesting unique position, if they play their cards right they could become a big player in the field, too many cards on AI might not be it though.

▲

reaperducer 7 hours ago | parent | prev [-]

I hope they don't bet too much on AI

Based on DO's monthly newsletters, all it cares about anymore is AI.

I don't think its written a single bullet point in two years that wasn't about AI.

It's as if some wave of amnesia came over DO and it forgot it has thousands of existing customers who are not AI wantrapreneurs.

▲

nine_k 7 hours ago | parent | next [-]

Come on, this is plain wrong, fortunately.

I'm subscribed to the DO newsletter; let me skim the archive. What they introduced in the last couple of years was: per-second billing, various identity providers for their SSO, managed Postgres upgrades, storage autoscaling, a NAT gateway, "bring your ow IP", etc. Yes, they do actively build out their GPU offerings, and access to open-weight models, but it's by far not the only thing.

	▲	giancarlostoro 6 hours ago \| parent \| next [-]
		There's a lot of AI fatigue to be fair, so its very easy to overlook the non-AI things too.
	▲	reaperducer 6 hours ago \| parent \| prev [-]
		Come on, this is plain wrong, fortunately. DigitalOcean newsletter for March 20,2026: - Deploy 2026: The Production Inference Era - DigitalOcean at NVIDIA GTC 2026: Building the AI Factory for the Agentic Era - The Proven Home for AI Agents - The Richmond Data Center: Our newest facility engineered exclusively for AI, featuring NVIDIA HGX B300 systems and a 400 Gbps non-blocking RDMA fabric. - Frictionless NVIDIA Partnership - Expanded Model Catalog - 1 click NemoClaw (Alpha) Droplet - Network File Storage standard tier and expanded availability 7 AI items. One non-AI item. I stand by my comment.

▲

gloryjulio 5 hours ago | parent | prev [-]

The problem is that their current capacity is literally full. They were running a highly profitable business for the last 2-3 years and recently switched up the strategy to build more datacenters to meet the ai demand.

I can see they want to do it as they are currently demand constrained.

	▲	gloryjulio 2 hours ago \| parent [-]
		Update typo: Meant to say supply constrained