Foxconn, the world’s largest electronics manufacturer, just posted the best first quarter in its history. And the reason has almost nothing to do with iPhones.
The Taiwanese contract manufacturer reported revenue of NT$1.64 trillion (approximately $53.4 billion) for the first three months of 2025, a 24.2% surge from the same period a year earlier, according to Yahoo Finance. March alone delivered NT$559.77 billion, up roughly 20% year-over-year. The company, formally known as Hon Hai Precision Industry Co., attributed the performance overwhelmingly to one thing: explosive demand for AI servers.
That single product category — AI servers and the networking equipment that supports them — has become Foxconn’s growth engine, eclipsing the consumer electronics business that defined the company for decades. The shift is not subtle. It’s structural. And it tells a broader story about where the real money flows when the technology industry commits to an infrastructure buildout of this magnitude.
From Assembly Floor to Data Center Backbone
For years, Foxconn was synonymous with smartphone assembly. The company built its empire on razor-thin margins and staggering scale, churning out hundreds of millions of Apple devices annually from sprawling factory complexes across China and, increasingly, India and Vietnam. That business isn’t going away. But it’s no longer the story.
The company’s cloud and networking products division — the segment that includes AI servers — has been growing at a pace that dwarfs everything else in the portfolio. Foxconn disclosed that this division was the primary driver behind the record Q1 figures, with demand from hyperscale data center operators showing no sign of cooling. The company builds servers powered by Nvidia’s GPU architectures, assembling the complex liquid-cooled rack systems that companies like Microsoft, Amazon, and Google are deploying at enormous scale to train and run large language models.
Chairman Young Liu told investors in recent months that AI server revenue could grow more than 40% in 2025. That projection, made before the Q1 numbers landed, now looks conservative.
Consider the context. Nvidia reported in February that data center revenue hit $35.6 billion in its fiscal fourth quarter, a figure that itself shattered records. Every one of those GPU shipments needs to be integrated into server systems. Foxconn sits at the center of that integration work, competing with Quanta Computer and Wistron but increasingly pulling ahead on volume and customer relationships. According to reporting by Reuters, the company’s AI server business now accounts for a rapidly expanding share of total revenue, a proportion that was negligible just three years ago.
The numbers are striking when you zoom out. Foxconn’s full-year 2024 revenue came in at NT$6.85 trillion, also a record, meaning the company is on pace to exceed that figure comfortably if current momentum holds. The stock has responded accordingly, rising more than 70% over the past twelve months on the Taiwan Stock Exchange.
But there are complications.
The most immediate is geopolitics. Foxconn’s manufacturing footprint remains heavily concentrated in China, and the escalating trade tensions between Washington and Beijing have introduced fresh uncertainty into supply chains that were already being restructured. The Trump administration’s tariff policies — including broad levies on Chinese-manufactured goods and specific restrictions on advanced semiconductor equipment — create real operational risk for a company that ships finished products across borders constantly. Foxconn has been diversifying production into India, Mexico, and Vietnam for several years, but relocating AI server manufacturing is far more complex than moving smartphone assembly lines. The servers require specialized components, advanced thermal management systems, and proximity to chip packaging facilities that remain concentrated in East Asia.
There’s also the question of margin. Foxconn has historically operated on net margins in the low single digits — a function of the contract manufacturing model, where the brand owner captures most of the value. AI servers carry better margins than consumer electronics, but Foxconn is still fundamentally a build-to-order operation. It doesn’t own the intellectual property in the GPUs, the networking chips, or the software stack. Nvidia, Broadcom, and the hyperscalers themselves capture the lion’s share of the economic surplus. Foxconn captures volume.
Volume, though, can be enormously valuable when the market is expanding this fast.
Capital expenditure plans across the major cloud providers reinforce the demand picture. Microsoft has signaled it will spend more than $80 billion on data center infrastructure in its current fiscal year. Meta has guided to $60–65 billion. Amazon’s AWS division is spending at a comparable clip. Google parent Alphabet has committed $75 billion. These aren’t aspirational figures. They’re budgeted commitments, and a meaningful share flows directly to server manufacturers like Foxconn.
The competitive dynamics within the AI server supply chain are also shifting. Foxconn has been investing in its own design capabilities, moving beyond pure assembly into system-level engineering for AI racks. The company has partnered directly with Nvidia on the design of GB200 NVL72 server racks — the next-generation liquid-cooled systems built around Nvidia’s Blackwell architecture. This kind of co-engineering relationship gives Foxconn a stickier position than traditional contract manufacturers typically enjoy. It’s harder to switch vendors when the vendor helped design the product.
Meanwhile, the broader Taiwanese supply chain continues to benefit disproportionately from the AI buildout. TSMC fabricates the chips. Foxconn assembles the servers. Delta Electronics supplies the power systems. Cooler Master and other thermal solution providers handle heat dissipation. Taiwan’s centrality to the AI hardware stack is, if anything, intensifying — a geopolitical fact that policymakers in Washington, Beijing, and Taipei are all acutely aware of.
Foxconn’s Q1 results also arrive amid a period of renewed anxiety about whether AI spending can sustain its current trajectory. The DeepSeek episode in January — when a Chinese AI lab demonstrated competitive model performance at a fraction of the typical training cost — briefly rattled markets and raised questions about whether hyperscalers were overbuilding. Foxconn’s numbers suggest the spending hasn’t slowed. Not yet.
So what comes next? The company has guided for Q2 revenue to be roughly flat or slightly higher sequentially, which would represent continued year-over-year growth given the weak comparable period in 2024. Management has emphasized that order visibility for AI servers extends well into the second half of the year, with some contracts stretching into 2026. The backlog, by all accounts, remains substantial.
There’s a broader lesson in Foxconn’s transformation. The companies that benefit most from a technology wave aren’t always the ones that grab headlines. Nvidia gets the magazine covers. OpenAI gets the cultural attention. But Foxconn — unglamorous, operationally intense, running on thin margins and massive throughput — is quietly converting the AI boom into record revenue quarter after quarter. It’s the pick-and-shovel play of the current cycle, updated for an era where the shovels cost millions of dollars and require liquid cooling.
The question for investors is whether this is a durable shift in Foxconn’s business mix or a cyclical peak that will fade when AI capital spending inevitably normalizes. History offers cautionary precedent. The company rode the smartphone boom to enormous scale, only to see growth flatten as the market matured. AI servers could follow a similar arc — torrid growth followed by commoditization and margin compression.
But for now, the trajectory is unmistakable. Foxconn isn’t just riding the AI wave. It’s building the physical infrastructure underneath it, one server rack at a time. And the first quarter of 2025 was its best ever.
WebProNews is an iEntry Publication