Foxconn's latest profit beat says the AI server boom is still showing up in factories, not just in Nvidia presentations and Wall Street forecasts.
Hon Hai Precision Industry, the Taiwanese company better known as Foxconn, reported second-quarter results on August 12, and the useful signal is plain: AI server orders are still running hot. The Wall Street Journal reported that Foxconn's net profit for the June quarter rose 35% to NT$59.97 billion, while revenue climbed to NT$2.526 trillion. That's a serious jump. If you've been waiting for proof that the AI infrastructure buildout is cooling, Foxconn didn't give it to you.
The company sits in a rare place in the chain. Nvidia can talk about demand. Cloud companies can talk about capital spending. Foxconn has to build the boxes. When the assembler's cloud and networking business becomes more than half of total revenue, as the Journal reported for the quarter, you are looking at physical demand moving through plants, components, racks and delivery schedules.
This isn't a one-quarter story. Reuters reported in May that Foxconn's first-quarter net profit rose 19% to NT$49.92 billion, beating the LSEG consensus estimate of NT$48.88 billion. First-quarter revenue reached NT$2.12 trillion, and earnings per share hit NT$3.56, according to the company's own release. Look at the mix. Foxconn said its cloud and networking product segment was then close to 50% of revenue, helped by AI server demand during what is normally a weaker season for the electronics industry.
That is the change.
The AI Rack Business Is Now The Business #
Foxconn is still Apple's top iPhone assembler, and that matters at holiday peaks. But the center of gravity has moved. Reuters described Foxconn as Nvidia's biggest server maker, and Hon Hai's own statements this year have repeatedly pointed to AI racks, cloud products and networking gear as the growth engine. In July, Reuters reported that Foxconn's April-to-June revenue rose 39.8% from a year earlier to NT$2.513 trillion, beating the NT$2.372 trillion LSEG SmartEstimate. June revenue alone hit NT$821.8 billion, a record for that month.
Those numbers are more useful than a clean narrative about an AI bubble. Bubbles can exist while factories are busy. Both can be true for a while. The point for you is narrower and more immediate: Foxconn's order flow still says customers are taking delivery of expensive AI systems at scale.
The company's guidance has been just as direct. At its May annual meeting in New Taipei City's Tucheng District, Taipei Times reported that chairman Young Liu set a target of NT$11 trillion in annual revenue for 2026, up 36% from last year. He also said Foxconn held more than 40% of the global AI server market and expected server rack shipments to more than double this year. That's not soft language. It is a production target tied to a giant procurement cycle.
Next-generation Nvidia systems keep that cycle moving. Nvidia said on May 31 that Vera Rubin was ramping into full production, with production shipments set to begin in the fall. Its list of manufacturers included Foxconn alongside Dell, HPE, Lenovo, Supermicro, ASUS, Gigabyte, Pegatron, QCT, Wistron and Wiwynn. TrendForce said in May that the top five North American cloud service providers would account for more than 60% of global demand for Nvidia GB and Vera Rubin series servers this year. That is the customer base Foxconn is chasing.
The Risk Is Concentration #
Here is the thing: strong demand doesn't make the business bulletproof. Foxconn's growth is now tied more tightly to a small group of cloud customers and Nvidia's platform cycles. It also depends on how fast data centers can absorb new racks. A by one large buyer would show up quickly. So would a delay in power or cooling at the data center level. So would slow construction.
Foxconn has already warned about the parts it can't control. Reuters reported in July that the company cautioned about a volatile global political and economic situation, even as it forecast third-quarter growth and said AI racks would keep their growth trend. That's the right caveat to keep in the story. This is not a smooth machine. It is a supply chain stretched across Taiwan, the United States, Mexico, Europe and a handful of hyperscale buyers with their own budgets and delays.
The broader market is watching the same signals. Nvidia has been telling investors there are hundreds of billions of dollars in AI infrastructure demand ahead, while TSMC reported a 77% jump in second-quarter profit in July, according to Reuters. Foxconn adds the manufacturing proof. Chips are being ordered. Racks are being assembled. Server revenue is replacing smartphones as the cleaner read on Hon Hai's future.
For now, it hasn't cracked. Frankly, if AI infrastructure spending were already rolling over, Foxconn would be one of the first places you'd see the dent. The latest quarter shows the opposite: profit above expectations, revenue at record scale, and an AI server business big enough to reshape one of the world's most important manufacturers.
Also read: A China-Based Hacker Let DeepSeek AI Run Cyberattacks on 460 Systems • Upstart Investors Sue Over an AI Model That Quietly Rejected Borrowers • Hetzner Quietly Launched a Free API for DeepSeek, GLM and Qwen Models