Big Manufacturers Find New Demand in Equipping AI Data Centers

For the past two years, if you asked anyone where the money in AI was being made, they'd point to one place: Nvidia. And they wouldn't be wrong. The chipmaker's GPUs have been the currency of the AI revolution, the scarce resource that everyone from OpenAI to your neighbor's startup has been scrambling to get their hands on.

But here's the thing about those GPUs, they don't do much without somewhere to live.

Someone has to build the servers that house them. Someone has to design the cooling systems that keep them from melting. Someone has to manufacture the power infrastructure that feeds the insatiable appetite of a modern AI data center. And that someone? A whole ecosystem of manufacturers that are suddenly finding themselves at the center of the biggest industrial buildout since the dot-com era.

This isn't just a chip story anymore. It's a manufacturing story. And it's rewriting the fortunes of companies you've probably heard of, and some you might not have.

The Scale of the Opportunity, By the Numbers

Let's start with some perspective. Because the numbers here are genuinely staggering.

A $1.8 Trillion Market in the Making

The worldwide Data Center IT Semiconductors and Components market is projected to reach $1.8 trillion by 2030, according to Dell'Oro Group. That's a revision upward from earlier forecasts, reflecting higher projections for data center capital expenditure and power capacity additions.

To put that in context: we're talking about a market that's reshaping entire industrial sectors.

The global AI data center infrastructure market alone is expected to reach $180.3 billion in 2026, growing at a compound annual growth rate of 35.5% to surpass $2 trillion by 2034. Meanwhile, the AI-ready data center infrastructure market is projected to grow from $28.4 billion in 2026 at a CAGR of 23.1%.

(Quick sidebar: If those numbers make your head spin, you're not alone. The scale of this buildout is genuinely unprecedented in the history of computing infrastructure.)

Record Orders and Backlogs

But let's move beyond forecasts to what's actually happening right now. The data center physical infrastructure market grew 28% year-over-year to $12 billion in manufacturer revenue in Q1 2026 alone, marking the fifth consecutive quarter of more than 20% growth.

Think about that. Five straight quarters of 20%+ growth. In an industry that, for decades, was characterized by steady but unspectacular single-digit expansion.

And the order books? They're overflowing.

Vertiv, a specialist in power and cooling infrastructure, reported a 252% surge in orders and a $15 billion backlog. Let that sink in for a moment. A quarter-billion percent increase in orders. The company's backlog alone is larger than many Fortune 500 companies' entire annual revenue.

Meanwhile, IDC projects AI infrastructure spending will reach $487 billion in 2026 and surpass $1 trillion by 2029. The global server market already hit a record $444 billion in 2025.

The Server Manufacturers Leading the Charge

So who's actually making all this stuff? Let's start with the most visible part of the ecosystem: the server manufacturers themselves.

Dell, From PC Maker to AI Powerhouse

Dell used to be the company you called when you needed a new laptop for the office. Not anymore.

In its first fiscal quarter of 2027 (ending May 1, 2026), Dell posted AI server revenue of $161 billion, a staggering 757% year-over-year increase. The company's AI server orders reached $244 billion for the quarter, and Dell raised its full-year AI server revenue expectation to approximately $600 billion.

Here's what's remarkable: Dell's Infrastructure Solutions Group operating profit hit $31 billion, up 206% year-over-year. That's not just growth, that's a fundamental transformation of the company's profit profile.

(The market noticed. Dell's stock jumped about 40% after hours when it reported these numbers.)

Supermicro, The Pure-Play AI Server Specialist

If Dell is the diversified giant pivoting hard into AI, Supermicro is the pure-play specialist that was built for this moment.

For the fourth quarter of fiscal 2026 (ending June 2026), Supermicro reported revenue of $111 billion, nearly double the $58 billion from the same quarter the previous year. Net income skyrocketed from $1.95 billion to $11.78 billion, an increase of over 500%.

The company's Q4 order book exceeded $600 billion, pushing backlog to record levels. And looking ahead, Supermicro expects fiscal 2027 net sales of $65-$72 billion, up from $39.1 billion in fiscal 2026.

AI GPU-related platforms now contribute over 80% of the company's revenue. This is a company that has bet everything on AI infrastructure, and so far, that bet is paying off spectacularly.

Lenovo, Infrastructure Business Doubles

You might know Lenovo for its ThinkPad laptops. But its infrastructure business is now the real story.

Lenovo's latest quarterly results showed its Infrastructure Solutions Group revenue reached $8.5 billion, up 98% year-over-year. The company's AI server pipeline stands at approximately $15.5 billion, with potential orders reaching $54 billion, a 150% sequential increase.

The company's infrastructure business is now profitable for the first time in years, and CEO Yang Yuanqing sees this as just the beginning.

The ODM Engine, Foxconn, Quanta, and Wistron

Behind the brand-name server makers lies an even bigger story: the contract manufacturers who actually build much of this hardware.

Foxconn (Hon Hai Technology Group) is reportedly seeing AI server rack shipments double in 2026, with Q3 shipments growing at a "high double-digit" percentage range. The company's GB300 rack shipments are expected to reach 15,000 to 20,000 units in the second half of 2026. In one of the most eye-popping deals of the year, SpaceX reportedly placed an order with Foxconn for approximately 13,000 Nvidia GB300 AI server racks, a deal valued at a potential $52 billion.

Quanta Computer, another major ODM player, has secured orders from Google, Microsoft, and Meta, with shipments expected to remain strong through the end of 2026. The company has also picked up orders from emerging "neocloud" customers, new cloud providers building infrastructure specifically for AI workloads.

And here's the kicker: ODM factories' AI server capacity for the second half of 2026 and 2027 is already largely booked. When your production capacity is sold out 18 months in advance, you know you're in a bull market.

Beyond the Server, The Physical Infrastructure Boom

But servers are only part of the equation. In fact, they might be the smaller part of the opportunity.

Power Infrastructure, The Hidden Bottleneck

Here's something that doesn't get enough attention: power is the real constraint on AI data center growth.

The International Energy Agency reported that data centers, AI, and cryptocurrency used about 460 terawatt-hours of electricity in 2022 and could exceed 1,000 TWh by 2026. Goldman Sachs projects global data center electricity demand will grow 50% by 2027 and up to 165% by 2030 compared to 2023 levels.

This creates massive demand for power infrastructure equipment, and a handful of manufacturers are perfectly positioned to capture it.

Eaton, a 112-year-old industrial conglomerate, reported Q2 2026 revenue of $8.53 billion, with data center organic revenue climbing 65%. The company's Electrical Americas segment posted record sales of $3.6 billion in Q1 2026. Eaton's cooling business alone is on track for $1.7 billion or better in full-year 2026 revenue.

Eaton provides heavy-duty transformers, switchgear, and uninterruptible power supply systems for megawatt-class server racks. Its next-generation 800-volt DC power distribution architecture is designed to streamline power delivery directly to servers, eliminating multiple conversion steps and significantly reducing energy loss.

Vertiv is even more focused on AI infrastructure. The company reported Q2 2026 net sales of $3.27 billion, up 24% year-over-year, with adjusted operating margin expanding to 22.6%. The Americas region grew 29% as hyperscale power and thermal orders compounded.

Vertiv is at the forefront of power architecture evolution, supporting both AC and emerging 800V DC solutions. The company has collaborated with NVIDIA and Foxconn's VisionBay AI on landmark projects, including Taiwan's first AI data center featuring NVIDIA GB300 and the world's first AI data center adopting 800V DC architectures.

The Cooling Revolution, Liquid Cooling Goes Mainstream

If power is the fuel, cooling is the insurance policy. And the insurance industry is booming.

Why Air Cooling No Longer Cuts It

Modern AI chips generate extraordinary amounts of heat. As AI chip power consumption and rack power density continue to climb, liquid cooling is transitioning from a nice-to-have to a must-have for high-density compute clusters.

The numbers tell the story: liquid cooling penetration in AI data centers is expected to reach 40% in 2026, up from 14% in 2024 and 33% in 2025. This represents a fundamental shift in how data centers are designed and built.

The global liquid cooling market is projected to reach $10.2 billion in 2026 and explode to $87.3 billion by 2030. That's not growth, that's a hockey stick.

The Liquid Cooling Supply Chain Emerges

A new supply chain is emerging to meet this demand. Companies like (Envicool), (Shenling) , and (Tongfei) are positioning themselves as full-system liquid cooling providers. Component suppliers like Feilong) and (Yidong) are also seeing increased demand.

Delta Electronics and Asia Vital Components are among the established players pivoting into liquid cooling. Asia Vital Components reportedly invested a decade in liquid cooling R&D before the market even materialized, a bet that's now paying off handsomely.

Super Micro Computer has expanded its Data Center Building Block Solutions portfolio with 10 precision-engineered rack models designed for high-density AI data centers. The company's global manufacturing operations can produce up to 3,000 advanced racks per month, including 2,000 liquid-cooled units.

Strategic Partnerships Reshaping the Industry

The scale of this buildout is so massive that no single company can do it alone. Strategic partnerships are forming across the ecosystem.

Schneider Electric and Foxconn

In June 2026, Schneider Electric announced a strategic collaboration with Foxconn to develop and scale next-generation AI data center infrastructure. The partnership focuses on co-developing reference architectures for AI data centers and exploring innovations in energy optimization, modular power, and cooling systems.

Production of the companies' first joint hardware solutions is expected to begin later this year. Schneider Electric's stock jumped 10% on the announcement.

Vertiv, NVIDIA, and the 800V DC Future

Vertiv is collaborating with NVIDIA on integrated infrastructure solutions, including 800V DC infrastructure designed for NVIDIA's Vera Rubin platform. The companies are also co-developing AI-native digital twin tools.

These partnerships reflect a broader trend: the lines between chipmakers, server manufacturers, and infrastructure providers are blurring. Everyone needs to work together to deliver the integrated solutions that hyperscale customers demand.

Global Manufacturing Expansion, Building Capacity for AI

The demand is so strong that manufacturers are scrambling to expand production capacity.

Foxconn, Quanta, and Wistron Lead the Charge

Taiwanese ODMs are accelerating overseas expansion to meet AI server demand. Foxconn, Quanta, Wistron, and Wiwynn are all investing heavily in new production facilities, particularly in North America.

The urgency is driven by record orders from hyperscale customers. The five largest North American cloud service providers, Google, AWS, Meta, Microsoft, and Oracle, are expected to invest nearly $900 billion in AI infrastructure in 2026, an increase of nearly 90% year-over-year.

Vertiv's Global Footprint Expansion

Vertiv has rapidly expanded its global manufacturing presence, adding new capacity in Malaysia, expanding five large plants in the Americas, and increasing chiller capacity in EMEA. These investments enable the company to deliver increasingly complex data center infrastructure solutions at scale.

What's Driving This Demand, And How Long Will It Last?

From Training to Inference, The Next Growth Wave

One of the most important shifts happening in AI is the transition from training to inference.

Lenovo CEO Yang Yuanqing noted that AI compute demand was previously about "80% training, 20% inference." That's now shifting toward "50% training, 50% inference," and the reasonable long-term state may be "80% inference, 20% training".

This matters because inference, the process of running AI models to generate outputs, tends to be more distributed and requires different infrastructure than training. It also represents a much larger addressable market in the long run.

As Lenovo's CFO noted, "enterprise-side AI inference demand hasn't really started at scale yet". When it does, it will drive another wave of equipment demand.

The CSP Spending Surge

The nine largest cloud service providers globally are expected to increase capital spending by approximately 90% in 2026. Chinese CSPs, ByteDance, Tencent, Alibaba, and Baidu, are collectively increasing spending by over 80%.

For 2027, total CSP capital spending is projected to reach $1.3 trillion. While the year-over-year growth rate will moderate due to the high base, spending will continue to set new records.

Challenges on the Horizon

It's not all smooth sailing. Several challenges could temper growth.

Power availability remains the defining constraint. "Access to power is still the defining challenge for the buildout," said Alex Cordovil, Research Director at Dell'Oro Group. Some projects are being delayed simply because there isn't enough grid capacity to power them.

Permitting delays and community opposition have emerged as real factors in project timing, particularly in the United States. Building a massive data center isn't just about technology, it's about navigating local politics and environmental regulations.

Supply chain bottlenecks persist. GPU availability remains constrained, and memory shortages are driving up costs. Component shortages across the supply chain could limit how fast manufacturers can scale.

The AI data center buildout is creating a manufacturing renaissance. From the server OEMs like Dell and Supermicro reporting triple-digit revenue growth, to infrastructure specialists like Vertiv and Eaton seeing orders surge 250%, to the contract manufacturers whose production capacity is sold out through 2027, the entire ecosystem is experiencing unprecedented demand.

And here's the thing: we're still early.

The transition from AI training to inference hasn't fully played out. Enterprise AI adoption is just beginning. The next generation of AI models will require even more powerful infrastructure. The global buildout of AI data centers is still in its early innings.

For manufacturers who can deliver the servers, power systems, cooling solutions, and components that AI data centers need, the opportunity is massive, and it's only going to grow.