The Great Neocloud Consolidation Has Already Started

Kevin Cochrane, Chief Marketing Officer at Vultr, discusses why the great neocloud consolidation has already started, and what it means for the market. This article originally appeared in Insight Jam, an enterprise IT community that enables human conversation on AI.

The cloud services market, once dominated by hyperscalers, is now crowded with alternative “neoclouds” that promise greater agility and lightweight solutions for enterprises seeking more efficient cloud infrastructure. But by the end of 2026, there will be far fewer neoclouds than there were at the start.

This year’s unique market conditions are set to precipitate significant consolidation. First-generation infrastructure is timing out. Enterprise contracts are up for reevaluation. GPU availability is expanding, but who gets them will be determined by strong supplier relationships and ample available capital. There is only so much funding to go around, and it will go to the companies that can vault the barriers of geographic reach, power capacity, hardware functionality, and regulatory scrutiny to be reliable partners to enterprises across the globe.

Charting the “Silicon Gold” Rush

Have land, have GPUs, will profit. 

Such was the mantra of the old crypto-mining days and an early driving force behind the exponential growth in GPU value. Then, the rise of generative AI gave GPUs a massive boost, creating the impression of a bulletproof business model: acquire a fleet of GPUs, rent them to a hardware-happy market, and reap the profits.

For a while, this worked. Around 2023-24, the market rewarded the minimum of basic GPU access. Data center operators sprinted into the neocloud business: They had the framework, so they invested in GPUs and began renting them out.

But not every company knew what to do with naked hardware. The hyperscalers were quick to assert their hegemony and, while some neoclouds succeeded, many others were stuck with unsustainable models. Now, with the 2024 cloud contracts signed and coming up for reevaluation, the cracks are starting to show.

As AI models become more complex and compute demand continues to skyrocket, the most influential customers—ones that can keep a cloud services business alive—need more than the newest chip on the block. Next-generation hardware must be bolstered by data sovereignty guarantees, latency optimization, ironclad protection of sensitive information, and unwavering operational resilience.

How a Neocloud Evaporates

In 2026, many neoclouds may simply run out of money.

Cloud operations is a highly capital-intensive business. Providers must raise capital at a volume and frequency that enable continuous growth, which means their customers must also be growing; infrastructure must be in place to support profitability and scale. Most neoclouds will need to double their customer base every 3 months to remain competitive.

This assumes they can afford the hardware. GPUs are expected to command the lion’s share of the $3.1 trillion in capital that will flow into computing hardware by 2030, and they can cost up to 3.7 times as much as hosting. A significant reserve of neocloud funding must go toward updated hardware. However, funding is increasingly difficult to acquire. With hundreds of cloud competitors, along with a host of AI startups, lenders will be choosy, and investors will be choosier. Cloud providers must continually build capacity, reinforce their value to customers, and increase their revenue to demonstrate the growth that attracts future investment.

Another major obstacle to neocloud survival is hardware depreciation. Any enterprise still running on 2021-22-era GPUs faces a complete fleet reconstruction. With GPU prices decreasing by half over about 5 years, cloud providers have less than 4 years to recover their capital and reinvest in the next generation of hardware. To sustain ballooning AI workloads, they’ll need to do this at an accelerated rate, while also maintaining support for customers on older hardware.

Even financially sound, GPU-rich cloud providers still have to stand up to regulatory pressure. Building the security and excellence to meet regulations takes a long time. Any entity that manages personal data will be subject to reviews of its hiring practices, training processes, and data governance. Fly-by-night neoclouds often lack the infrastructure or manpower to meet such stringent compliance benchmarks.

The most successful neoclouds will be those that not only stand up to scrutiny but also prove themselves valuable partners to the nations where they operate. Regulatory bodies and national governments don’t want to contend with hundreds of providers to ensure their citizens’ data is being managed within compliance. Instead, they’re throwing their weight behind “national champion” providers that they can trust.

2026 is an inflection point. Only a handful of neoclouds will survive the pressure. Even fewer will become market leaders. What happens in this decisive year will reshape the market underlying the biggest technological shift of this century.

The Formation of a Thriving Neocloud

The difference between a neocloud that’s barely surviving and one that will thrive into 2027 and beyond is whether they can accommodate the enterprise AI shift. As AI exits the experimentation phase and begins to transform business operations at scale, the standard for a suitable cloud introduces several new criteria. GPU access is the bare minimum. Today’s enterprises need proof of resilience at every layer of the stack to feel confident in choosing a service provider.

The rapid evolution of agentic AI applies even greater pressure. Agentic workloads require massive compute power and storage capacity. They also require increased visibility, as humans still need to track autonomous execution, and high security standards for models utilizing sensitive data. Three critical capabilities will determine which neocloud companies thrive on their own and which become acquisition fodder.

The first is capital access. A few large enterprise contracts can prop up a neocloud in the short term, but remove one load-bearing customer, and the whole operation could topple. A truly stable neocloud will generate revenue across a broad customer base, from small and midsize organizations to major enterprises, ensuring ample cushion to invest in next-generation silicon.

This is why delivering more than just GPU access is so critical: The combination of a functional stack with silicon diversity creates more avenues to deliver value, thereby attracting customers across more sectors. Greater industry penetration then unlocks new revenue opportunities. Multilayered infrastructure is what best fits today’s emerging AI use cases, and few neoclouds will be able to provide it without consolidating.

The second critical capability is operational scale. Successful neoclouds will have a broad geographic reach, enabling them to serve more customers across more countries and, in turn, empowering those customers to achieve global growth.

Geographic reach will also be crucial as cloud sovereignty gains momentum. While AI is maturing in the private sector, a similar shift is underway across governments and public institutions; countries are aligning on their AI priorities, a process that’s unearthing a widespread need for enhanced digital sovereignty.

Operational scale also encompasses power capacity. It doesn’t matter how many GPUs a provider has if they don’t have the power to run them. Residence in data centers with efficient cooling and power systems guarantees consistent capacity for advanced workloads. There’s also a market bonus: amid rising concern about the ecological impacts of AI, energy-efficient operations are a powerful draw.

From here, it’s not about having GPUs but about operationalizing them. Providers must build out the networking, burn in the GPUs, onboard multiple customers’ workloads onto the GPUs, and then maintain and grow their customer base by delivering constant uptime and exemplary service. Very few neoclouds can do this within a single data center, let alone across multiple locations.

The third key capability is go-to-market strength. Sustaining a cloud operations business requires far more than simply renting hardware. The real power comes from who is renting it.

Cloud businesses must have a strong acumen for enterprise sales – individual developers can be great customers if you have a small army of them, but enterprise contracts are what keep the lights on. The neoclouds that survive will be those that attract a broad portfolio of customers across industries. These providers will operate more like alternative hyperscalers than true neoclouds, enabling customers to maximize compute power, security, residency, and accessibility across their workloads–without blowing their budgets.

That means not necessarily competing with the hyperscalers, but functioning as their strategic complements. Many large enterprises still depend on hyperscalers for certain features. Moreover, many advanced AI projects are best optimized in a multi-cloud environment. With cloud services shifting toward an open ecosystem, walled gardens are bound to wither.

Forecasting a Condensed Cloud Market 

By 2027, a slim handful of GPU providers will control 80 percent of the market. Consolidation won’t happen all at once, but the process is already in motion. Enterprises that rely on neoclouds should keep a close eye on their providers in case they need to migrate their services.

Ultimately, consolidation is a sign of a maturing market. A smaller field of competitors lessens the overwhelm for enterprises and public institutions making critical infrastructure decisions. Their cloud options will be bolstered by a baseline level of security and reliability. At the same time, there remains enough healthy competition in the market to drive innovation, leading to better solutions for all.