Foxconn Technology Group, a Taiwanese company best known for putting together iPhones, just had its best first quarter ever. The company’s revenue for the first three months of the year rose 24.2% year over year to NT$1.64 trillion, which is about $53.5 billion. This was mostly due to the huge AI servers demand. The outcome was more than just good. It was a signal flare for the whole world supply chain.
Yahoo Finance’s numbers confirmed what industry experts have been saying for months – Foxconn’s shift toward AI infrastructure is paying off in a way that dwarfs its traditional consumer electronics business. Foxconn said that the company’s cloud and networking products segment, which includes artificial intelligence servers, saw significant growth and became the biggest source of revenue for the quarter.
This company makes about 70% of the world’s iPhones.
Chairman Young Liu was very clear during the company’s earnings call. He said that 2025 would be a year of strong growth and that Foxconn expects its AI server sales to be higher than the total sales of all its other products.
That’s an incredible thing for a company to say that makes more than $200 billion a year. Liu has been getting Foxconn ready for this moment since at least 2023, when the company started rapidly increasing its server manufacturing capacity in Mexico, the US, and Taiwan.
It’s not a coincidence that the time is right. The amount of money spent on AI infrastructure around the world has gone through the roof. Microsoft, Amazon, Google, and Meta are all spending hundreds of billions of dollars on building data centers and filling them with artificial intelligence servers that are full of GPUs.
Nvidia, whose chips power most of these systems, had record data center revenue in the last quarter. Foxconn is at the center of this spending boom because it is the main contract manufacturer for Nvidia’s GB200 and GB300 server racks, which are some of the most complicated and expensive computer systems ever made in large quantities.
But Foxconn’s success in the first quarter didn’t happen in a vacuum. It happened during one of the most unstable periods for trade policy in a long time.
The Trump administration’s harsh tariff policy, which got much worse in April 2025 with broad reciprocal tariffs on Chinese goods and targeted taxes on electronics imports, has thrown global manufacturing into a state of deep uncertainty. Taiwan has to pay a 32% reciprocal tariff, but a 90-day pause that was announced in April brought the effective rate down to 10% for most goods. Semiconductors and some electronic parts were given special exemptions, but the bigger picture is still unclear. Foxconn said in its earnings report that trade policy is still a risk, but the company has been spreading out its manufacturing operations for years to reduce this risk.
That diversification is real and getting faster. Foxconn has promised to spend about $1 billion to grow its business in India, where it makes iPhones for both the Indian market and exports. The factory complex in Sriperumbudur, which is close to Chennai, now employs tens of thousands of people. The company has big facilities in Ciudad Juárez and Chihuahua, Mexico, where they put together servers and other products that will be sold in North America.
In Wisconsin, the Mount Pleasant campus, which has had a lot of problems in the past and was once called a political boondoggle, is now quietly switching to making data center servers instead of the LCD panels it was supposed to make.
The geographic spread is very important right now. After the most recent round of increases, companies that make most of their goods in China have to pay high tariffs that can be more than 145% on some items. Foxconn still has huge factories in mainland China, especially in Zhengzhou and Shenzhen. However, it can move production to other places, which gives it an advantage over smaller competitors.
What does the competitive landscape look like?
Pegatron is Foxconn’s closest competitor in contract electronics manufacturing. It also reported first-quarter results that showed growth, but at a slower rate. Wistron and Quanta Computer, two big names in server manufacturing, have also benefited from the growth of AI infrastructure. But none are as big as Foxconn. The company’s size – over 800,000 people work there around the world, and hence lets it deal with problems in the supply chain, get good prices on parts, and speed up production at rates that smaller companies can’t handle.
Nvidia has had a huge impact on Foxconn’s path. The two companies have worked together much more closely in the last 18 months. Foxconn is a major assembler of Nvidia’s DGX and HGX server platforms, and it has been chosen as a key manufacturing partner for the new Blackwell Ultra and Rubin architectures.
Jensen Huang, the CEO of Nvidia, has publicly praised Foxconn’s manufacturing skills many times, including when they both spoke at Foxconn’s annual tech day in October 2024. Foxconn needs Nvidia’s market dominance in AI accelerators so as to keep its server revenue growth going, and Nvidia needs Foxconn’s production power.
This partnership has a financial side that is worth looking into. Artificial intelligence servers cost a lot more than regular business servers. A single Nvidia GB200 NVL72 rack can cost more than $3 million. People think that the profit margins on putting these systems together are higher than those on consumer electronics, but Foxconn doesn’t give enough information about how much profit each segment makes to be sure of this. Morgan Stanley and JPMorgan analysts think that Foxconn’s AI server gross margins could be 200 to 300 basis points higher than the company’s average, which was around 6.5% in the last few quarters.
At Foxconn’s level of sales, even small improvements in margins are important. A 1% increase in margin on $200 billion in annual revenue means $2 billion more in gross profit. That’s the math that makes investors excited.
Over the past year, Foxconn’s shares on the Taiwan Stock Exchange have gone up about 85%, which is much better than the TAIEX index as a whole.
Things don’t always go smoothly. The consumer electronics part of Foxconn’s business, which is still a huge part of it, saw revenue stay the same or go down slightly in the first quarter. Demand for iPhones is steady, but it hasn’t grown as quickly as it did in previous upgrade cycles. Apple’s own AI goals, which are part of the Apple Intelligence brand, haven’t yet started a hardware refresh cycle that would significantly increase the number of handsets sold. Foxconn’s parts and computing products divisions also had mixed results, with some sub-segments seeing a drop in sales from the previous year.
Then there’s the tariff wild card, and there is no guarantee that it will be extended. If the full 32% tax on Taiwanese goods goes into effect, it could make Foxconn’s products that go to the U.S. a lot more expensive, even with the company making more products in different places. There have been reports that the company has been talking to customers about ways to share costs, but the details are still private. According to sources in the industry, hyperscale cloud providers are willing to take on some tariff-related cost increases instead of delaying server deployments because they need to keep their AI infrastructure buildout timelines. That makes Foxconn’s pricing power strong for now.
Chairman Liu talked directly about the tariffs, saying that Foxconn has multiple scenarios planned and that the company’s global manufacturing network provides natural hedging against trade problems. He didn’t give specific financial advice based on the outcome of the tariffs, which was a smart move because U.S. trade policy is so unpredictable right now.
The bigger AI server market doesn’t look like it’s going to cool down. TrendForce estimates that shipments of artificial intelligence servers around the world will grow by more than 40% in 2025, bringing the total market value to over $150 billion. Foxconn is thought to have about 30–35% of this market, making it the biggest company by revenue. Reports say that the company’s backlog of orders for artificial intelligence servers goes well into 2026. This gives the contract manufacturer a unique view of its future revenue, since it usually works with shorter planning horizons.
One thing to keep an eye on is Foxconn’s move into sovereign AI infrastructure. Several countries, such as Saudi Arabia, the UAE, Japan, and Indonesia, have said they will build huge data centers as part of their national AI computing plans. Foxconn has made itself a one-stop shop for these projects by offering not only server assembly but also data center design, construction management, as well as ongoing maintenance. This is a big step up from regular contract manufacturing, and it could lead to higher-margin revenue streams over time.
The company’s plans for electric cars, which are still in the early stages, add another layer. Foxconn’s MIH open EV platform has gotten a lot of automakers to work with it, and in late 2024, the company started making a small number of Model C crossover SUVs in Taiwan. But the money made from EVs is still very small compared to the server and electronics businesses. Liu has called EVs a long-term strategic investment instead of a short-term growth driver. This is a realistic view because the automotive industry is very competitive and needs a lot of money to grow.
Back to the primary story. Foxconn’s first-quarter results show a change that has been happening for years. A company that has long been known for putting together Apple’s hardware is quickly becoming the backbone of AI infrastructure around the world. The mix of revenue tells the story – for the first time on an annual run-rate basis, cloud and networking products are Foxconn’s biggest business, beating out consumer electronics.
There are risks with that change. Foxconn’s fortunes are closely tied to the capital spending cycles of a few hyperscale customers because it makes a lot of AI servers. Foxconn would feel the effects strongly if AI investment slows down because of a recession, a re-evaluation of AI’s short-term commercial returns, or a technological disruption that lowers demand for current-generation hardware. The company’s margins are pretty thin, so it doesn’t have much room to deal with drops in revenue without cutting profits significantly.
But for now, the AI Servers demand signals are mostly good. Microsoft has promised to spend more than $80 billion on data centers that can handle AI in fiscal year 2025. Meta’s capital spending plans for the year are between $60 billion and $65 billion, with most of that going toward AI infrastructure. Amazon and Google have also said they will spend a lot of funds. These are not guesses about what will happen. They are committed budgets that are backed by signed purchase orders, many of which go straight to Foxconn’s factory floors.
The first quarter of 2025 was a proof point. Foxconn is more than just a part of the AI infrastructure boom. It’s making it.