The AI industry continues to advance in computing power, models, and applications, with fundamentals steadily improving. Despite short-term market fluctuations, leading companies are strengthening industrial support through investment, capacity expansion, and technology implementation. AI is transitioning from expectations to substantive commercialization, maintaining a positive mid-term outlook. Looking ahead, three key investment themes are recommended: 1) Large model and agent platform expansion; 2) Continued capital expenditure in infrastructure chains such as AI servers, power, and cooling; 3) Mass adoption of consumer-facing AI applications and embodied intelligence.
This week, the global AI industry remains highly active, with accelerated progress in computing infrastructure and productization of large models. The sector exhibits a dual-driven trend of "technological leaps + ecosystem expansion." Google officially launched its next-generation flagship model, Gemini 3, which was integrated into Search, the Gemini App, and developer platforms on the same day—far exceeding market expectations. Gemini 3 topped LMArena benchmarks, significantly enhanced reasoning capabilities, and introduced Deep Think mode alongside the agent development platform Antigravity, further solidifying Google's competitive edge in multimodal and reasoning domains. Market response to the update has been positive, with Google's stock showing steady gains, reflecting strengthened long-term competitiveness through model iteration.
AI infrastructure maintains strong momentum. Foxconn announced a collaboration with OpenAI to build next-gen AI data center racks in the U.S., covering critical components like power, cooling, and networking to bolster North America's computing supply chain. Nokia, leveraging its partnership with NVIDIA, unveiled a new organizational and strategic framework focused on AI-RAN, data center networks, and next-gen communication facilities, accelerating its transformation into an "AI-driven communication infrastructure provider." These infrastructure initiatives underscore industry leaders' consensus on the AI supercycle: computing demand will remain robust long-term, with data centers, GPUs, and power equipment continuing to attract significant global capital expenditure.
Innovation persists on the model and application front. Google introduced Nano Banana Pro, a next-gen image generation model based on Gemini 3, featuring upgrades in multi-reference consistency, text-controlled generation, and professional-grade detail—enhancing tools for creative design and marketing. Alibaba officially launched its "Qianwen" project and app, marking its full entry into the consumer AI market and intensifying competition among major tech firms in agent ecosystems. The trend toward platform-based agents is evident, as seen with Google’s Antigravity and Alibaba’s Qianwen, signaling AI's evolution from single-point interactions to multi-step, action-capable systems.
Comments