The State of Prism: Q1 2026 Update

Hey everyone, the team at Prism has been spending a lot of time recently reflecting on how we can continue pushing the bar, and we've decided to bring back our transparency reports. We used to put these out roughly every month, and while we loved doing that, the cadence wasn't always realistic to sustain. Going forward, we're making these a quarterly tradition. Less frequent than before, but consistent, and frankly better suited to delivering updates of real substance rather than scrambling to fill a page every few weeks. Most hosts don't do this at all. We believe deeply in transparency and in cultivating a genuine relationship with our clients, so here we are again, and we have a lot to talk about. Since this is our first one back, we're effectively covering the past year of Prism in its entirety, so grab a drink and settle in.

Growth

As many of you have no doubt noticed, over the past six months we've experienced what can only be described as extraordinary growth, nearly tripling our Discord member count (we just crossed 3,000) and sustaining roughly 34% month-over-month expansion. I genuinely cannot believe I'm writing this, but we now have over 100 machines in active service and have brought on our first full-time team member. Growth like this does not happen in a vacuum. It is the direct result of every single one of you who chose Prism, who stayed with us through the inevitable growing pains, who recommended us to a friend, or who simply showed up in the Discord and made this community what it is. Every server order, every support ticket (yes, even those), every candid piece of feedback has left its mark on what Prism is today. This team pours an unreasonable amount of heart and energy into building something we're proud of, and knowing that it genuinely resonates with the people using it is what keeps us moving forward. From the bottom of our hearts, thank you.

Infrastructure & the Memory Crisis

With growth, inevitably, comes challenge. As you're likely aware (we've touched on this a few times), we are currently navigating a global memory supply shortage that has driven our RAM procurement costs up by nearly 500%. RAM is arguably the single most critical resource for Minecraft servers after CPU compute (which remains in healthy abundance), and it simply cannot be substituted without meaningful performance degradation. Given that this crisis is fundamentally temporary and is expected to persist for another 1-2 years (or sooner if the AI bubble pops), it does not make economic sense to commission new builds at current prices.

Some context for those who are newer to Prism: we began as a developer-first platform, purpose-built to deliver the absolute fastest servers on the market with the highest reliability at the most competitive price. While we've since broadened well beyond developer use cases, those founding principles shaped our infrastructure philosophy from day one. We deployed exclusively on infrastructure we owned or on that of robust partners, but for the most part we had our own datacenter colocation contracts, our own network transit agreements, our own physical machines/network gear. This afforded us end-to-end command over the entire stack. We selected our own network transit providers, managed our own BGP announcements, operated our own inline DDoS mitigation pipeline, and addressed hardware failures directly on the datacenter floor. That depth of control is precisely what enabled the performance consistency and rapid incident response that distinguished us early on.

The tradeoff, of course, was that scaling under this model has always been demanding. We hedge hardware purchases, hold unused capacity in anticipation of demand, and absorb the capital risk that comes with it. We are still actively doing this. That said, we are candid about the fact that we're uncertain how much longer this approach remains viable at current memory prices. For now, it works, but it is something we are monitoring very closely.

Rather than passing unsustainable costs onto our clients (we would need to price servers well above $8/GB just to break even on new builds assuming we maintain our current ROIs), we have expanded into sourcing from established dedicated server providers and treating that layer as infrastructure to be abstracted upon. In practice, this means we overlay our own virtual network, retain our own BGP announcements and IP allocations, and deploy our full software stack on top, such that the client-facing experience remains identical. Your server's IP, our network, our panel, our support. We are essentially allowing upstream providers to absorb the hardware pricing impact across their longer investment horizons, taking a measured hit to our own margins rather than inflating yours. This is how we have been restocking popular locations in recent months, and clients should observe no difference in service quality whatsoever.

It is not a perfect arrangement, and we unfortunately remain bound by the laws of physics (and economics), but it allows us to continue scaling without compromising on the things that matter.

Pricing Changes

We do not anticipate any further price increases at this time. That said, we are making targeted adjustments to our tier structure to better reflect the current cost landscape and deliver a stronger experience across the board.

New Premium+ Tier

We are introducing a new flagship tier at $3/GB on Ryzen 9950X hardware. Think of it as an elevated iteration of our current Premium offering, designed for those who demand the absolute best silicon available. The existing Premium tier remains at $2.50/GB and will be first in line for Ryzen 11950X (or whatever Ryzen AI garbage they may rename it to) hardware soon after those chips become available, which is rapidly approaching.

Revamped Basic Tier

As we announced on March 13th, we have discontinued new orders on our original Basic lineup. That tier was conceived as a specialized offering for advanced users comfortable with hands-on resource management, but in practice the lower price point consistently attracted users expecting a standard plug-and-play experience, creating a persistent mismatch in performance expectations. Coupled with the reality that sustaining that price point during a memory shortage was no longer tenable without sacrificing quality, retiring it was the right call. If you are an existing Basic customer, your server remains fully active and supported, nothing changes for you.

In its place, we are launching a revamped Basic tier at $2/GB with more generous default allocations on previous-generation hardware. Stock for this tier will not be prioritized, so availability may be limited at times, but the intent is clear: a genuinely good entry-level experience that works out of the box without requiring you to manually configure processing threads and JVM flags. Better defaults, a better experience, and an accessible price point.

Stability & Network

It has been a very interesting year on the stability front, to put it diplomatically. The Minecraft hosting industry experienced unprecedented DDoS activity over the past twelve months, affecting virtually every provider in the space. We were not immune, and we have no interest in pretending otherwise. There were periods where attack volume and sophistication exceeded what our mitigation infrastructure at the time was architected to withstand, and some of you bore the consequences of that. We are sincerely sorry.

What we have done since: we undertook a comprehensive overhaul of our network and mitigation architecture. We have largely migrated our fleet to DataPacket with additional hardened protective measures layered on top, significantly diversified our upstream transit, and rebuilt our filtering pipeline from the ground up. The result is a materially more resilient network than what existed six months ago. We will not stand here and claim invincibility (no one credibly can), but we are in an incomparably stronger position than we were, and we continue to invest heavily in this area.

Panel & Software

The panel has received a substantial number of new features over the past year, which I will spare you the exhaustive list of here (and yes, we incorporated some AI capabilities... you're welcome, or we're sorry, depending entirely on who you ask). The more significant news is that we are approaching the launch of PrismPanel v2: a ground-up overhaul featuring a completely redesigned layout, refined UI throughout, and a meaningfully elevated user experience across every surface. We are genuinely excited about this one and will have much more to share in the near future.

Support

Let's be honest with you: our support infrastructure has not scaled proportionally with our growth, and that is unacceptable to us. When your user base triples in six months, certain seams are exposed, and support is where that strain has been most acutely felt. We are fully aware of this, and it is a top priority moving forward. We are actively in the process of expanding and strengthening our support team to ensure that the experience of getting help is worthy of the product itself. We will have more concrete updates on this soon, but we felt it was important to name this directly rather than gloss over it.

What's Next

We are heads down on several fronts: the Panel v2 launch, continued network fortification, support team expansion, and exploration of new locations as provider stock permits. We will keep you informed as these efforts progress, and you can expect the next edition of this report in Q2.

That said, these reports are not meant to be a one-way street. If there is something you think we should be prioritizing, something we've overlooked, or something you'd simply like to see more of, we genuinely want to hear it. Drop your thoughts in our feedback channel or open a ticket. The best decisions we've made have come from listening to the people actually using what we build, and that isn't going to change.

As always, thank you for being here. Prism exists because you chose it, and we do not take that lightly.

• The Prism team