📌 AI Chip Infrastructure Earnings Still Need a Close Look

BullSignal Automated Editorial System Published

Semiconductor stocks have come under pressure recently, but these earnings details show that orders and revenue for AI-related infrastructure have not weakened in sync. More importantly, the strong earnings come from different links: Cadence represents chip design software, while Amkor represents advanced packaging and testing. The former raised revenue and guidance, while the latter’s revenue, gross margin, and next-quarter guidance were all better than previous expectations, which has more reference value than a rebound in a single company.

Cadence’s change is relatively direct. QJun26 revenue grew 7.5% sequentially and 24.2% year over year, and the company raised its FY26 revenue guidance for the second time. The characteristic of EDA software is that it serves chip design activity, and an upward revision to revenue guidance shows that customers’ design investment is still continuing. For investors, this type of company does not need to build factories on a large scale like the manufacturing and packaging links, so revenue growth is more likely to be reflected in profit quality.

Amkor’s situation is more complex. 2Q26 revenue and gross margin were above the upper end of guidance, and 3Q26 continues to give a higher revenue and gross margin range, indicating that advanced packaging demand and capacity utilization are improving. Some analysis points out that over the past 18 months, Amkor’s customer concentration has declined, the share of advanced products has increased, and utilization has risen from 50% to 78%; all of these can amplify earnings leverage.

But Amkor also maintained FY26 capital expenditures of USD 2.5 billion to USD 3 billion, of which 65% to 70% will be used for facility expansion. The key here is not “whether the boom exists,” but “whether the boom can cover the cost of expansion.” If new capacity is absorbed smoothly, revenue and gross margin will continue to benefit; if the pace of demand is slower than expansion, depreciation, cash-flow pressure, and the return cycle will all affect shareholder returns.

So this set of information is more suitable for reaching a restrained conclusion: AI semiconductor infrastructure still has fundamental support, but industry prosperity cannot be directly equated with stock returns. Next, it is necessary to continue watching whether orders can be converted into utilization and free cash flow, rather than only looking at the scale of capital expenditures itself.

Sources