When Windrose launched into early access in April 2026, it became an instant hit.
Developed by indie studio Kraken Express, the co-op survival game sold 500,000 copies within 48 hours and reached a peak of 222,000 concurrent players.
Typically, it’s AAA titles that drive massive increases in server demand, but Windrose’s success reflects a broader industry trend where independent games are now also contributing significantly to these spikes.
Raphael Stange, the CEO of Nitrado – which supported the launch of Windrose with its multiplayer server platform GameFabric – explains that a decade ago, the game server hosting market was dominated by AAA shooters, MMOs, and the first wave of survival games such as Rust, DayZ, and Ark.
“What’s changed since then is scale and structure. Indie titles now routinely produce launches with AAA-level concurrent player counts. The hosting demand they generate is structurally different from what AAA shooters need: longer tails, more community-driven, more dependent on private server infrastructure than on official matchmaking. Minecraft was the early signal of this, a single-developer project that ended up generating more private server demand than most studio titles of its time.”
And there’s data to back this up. According to Newzoo, the share of PC playtime captured by games outside the Top 20 grew from 33% in 2022 to 42% last year. In Western markets, over half of PC gaming revenue now sits below the Top 20, up 48% from three years earlier.
Stange highlights the success of Palworld, which reached 2.1 concurrent players in January 2024, marking the third-highest peak in Steam’s history. This exceeds most AAA launches, he says, highlighting how the nature of player expectation and server demand has fundamentally changed.
“The communities around survival and sandbox titles are intensely engaged. They mod, they build, they run servers for dozens of friends. That generates a different kind of long-tail server demand than a passive player base does. A community that sets up a server for Ark or DayZ often runs it far longer than the typical hosting timeline. That tail is the part of the market most people underestimate.”
Indie infrastructure
Nitrado monitors upcoming titles from early in their development to determine the level of server infrastructure they might require.
“The pattern that gets our attention is consistent across games that turn into hits: multiplayer-first design, a genre with proven community server demand, and a team that understands the community they’re building for,” says Stange. “Windrose fit all three.”
Nitrado first engaged with Kraken Express in early 2025, when Windrose was known as Crosswind and had about 8,000 Steam followers, and the companies signed a partnership agreement five months before launch.
Stange says Nitrado and Kraken Express “knew the potential of Windrose was significant,” but its launch success surpassed expectations.
“What that moment actually tests is not whether your servers stay up under load. That’s the minimum. It’s whether the relationship is mature enough to make real-time decisions: capacity reallocation, region balancing, and rapid response to issues the community surfaces in the first 48 hours. Joint testing done in advance, shared playbooks, direct lines between engineering teams on both sides. That part of the work doesn’t show up in a pitch deck, and it’s the part that determines whether a launch goes well or badly.”
“The pattern that gets our attention is consistency: multiplayer-first design, a genre with proven community server demand, and a team that understands their community”
Stange said that a popular demo or beta is a sure sign of success, but it’s actually the subtle signals that are more useful when predicting a hit.
A community naturally taking shape is a notable example, with Discord servers forming, guilds being planned, and server configurations being debated months before launch. “That’s a signal that the design has created ownership, and ownership translates into sustained play.”
Another is when a genre has pent-up demand. But for Stange, the most beneficial signal is “how a studio thinks about infrastructure in pre-production.”
“Studios that come early and ask detailed questions about scaling behaviour, about what happens at 10x peak, about regional latency tradeoffs, those conversations correlate with games that ship well. It’s a proxy for how the team thinks about risk.”
“None of these guarantees anything. I’ve seen games where every signal pointed up and the launch underperformed, and games that weren’t on our radar much at all that broke out. The honest position is that these signals shift the odds; they don’t predict outcomes. Our job as a partner is to be ready for the upside scenario, not to claim we can see it coming.”
Unexpected demand
Nitrado also provided server support for Ark: Survival Evolved from Studio Wildcard, which launched into early access in June 2015 with a team of about 15 people.
“The sustained community server demand from that player base went far beyond what anyone, including us, had sized initially,” says Stange. “The most engaged communities ran their servers far longer than our typical hosting timeline, and that changed how we define a successful launch from an infrastructure perspective: the first weekend matters, but the multi-year tail matters more.”
He says Nitrado was underprepared for Palworld’s 2.1 million concurrent users during its launch. “That’s a number that breaks pure-cloud cost models and pure-bare-metal capacity models simultaneously.” (A bare metal server, or dedicated server, is where a computer is dedicated to a single customer, whereas cloud servers involve the use of multiple virtual machines, and can be scaled rapidly.)
“The studios that navigated those moments best had hybrid architecture in place before launch: predictable load on bare metal for cost efficiency, cloud capacity ready to absorb the spike. Pure cloud at that scale becomes financially unsustainable within days. Pure bare metal cannot respond fast enough to a demand curve that doubles in 48 hours.”
Stange said structuring infrastructure is most important for small studios scaling servers when a game is a hit, rather than an absolute budget moment.
“A bad launch for a small studio isn’t a bad weekend; it permanently damages community trust at the moment you most need it”
“Cloud-only is appealing because upfront commitment is low. But cloud pricing at viral scale, during a launch where concurrent count doubles every 48 hours, produces invoices that no small studio modelled in pre-production could afford. Bare-metal-only is cost-effective for predictable load, but cannot respond to a demand curve that breaks the ceiling.”
“The hybrid model is what works at scale: bare metal for the baseline you’re confident in, cloud capacity ready to absorb the spike, integrated DDoS protection because viral games attract attackers within hours of going live, and observability that lets engineering teams see what’s happening in real time. The ability to provision capacity in minutes rather than hours is the difference between a manageable launch and one that the community remembers for the wrong reasons.
“My direct advice to any small developer entering Early Access: have this conversation before you finalise launch budget, not after wishlists start moving. A bad launch for a small studio isn’t a bad weekend; it permanently damages community trust at the moment you most need it. Some studios don’t recover from that.
“The point is that the hard conversation about scaling happens once, in advance, instead of during launch weekend.”
