Chip Earnings Test Meets China's Open-Model Push

Two signals from AI hardware expectations and China's open-model challenge.

🗄️ Chip earnings face a high bar

Decoded: Reuters reported July 20 that semiconductor investors are looking for the sector to deliver nearly half of second-quarter S&P 500 earnings growth, with S&P 500 semiconductor and equipment earnings forecast to rise 133% from a year earlier, according to LSEG. The PHLX Semiconductor index is still up 65% for the year, but Reuters said it had fallen more than 20% from its late-June closing high and was down 18% in July. Intel and Texas Instruments report this week, while Nvidia reports in late August. Source: https://www.reuters.com/markets/us/chipmakers-head-big-profit-gains-will-it-be-enough-2026-07-20/

Why it matters: AI hardware is priced for strong execution, not merely strong demand. If INTC, TXN and peers disappoint on outlooks, investors may treat the weakness as a read-through for the whole chip basket, including TSM and AVGO.

🤖 Chinese open models raise pricing pressure

Decoded: The Verge reported July 20 that Moonshot AI unveiled Kimi K3 and Alibaba previewed Qwen3.8, with both companies claiming performance near top OpenAI and Anthropic systems. Moonshot describes Kimi K3 as a 2.8 trillion-parameter open-source model and says full weights are due July 27, while Alibaba says Qwen3.8 is a 2.4 trillion-parameter model going open-weight soon. Source: https://www.theverge.com/ai-artificial-intelligence/967781/chinese-ai-models-open-source-moonshot-kimi-k3-alibaba-qwen

Why it matters: The investor question is whether frontier capability stays scarce enough to justify closed-model pricing and US data-center spending. If capable open models from China keep narrowing the gap, cloud buyers gain leverage and BABA gets another proof point that AI distribution can matter as much as raw chip access.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team