CONNECT WITH US
Dec 31
Nvidia's robotics chief says the industry is building the wrong brains
Although AI has made tremendous progress in the digital domain, intelligence in the physical world still faces many challenges. Robots need to perceive 3D space, manipulate objects, and understand physical rules, all of which require enormous investments of manpower and resources. Dr. Jim Fan, head of Nvidia's robotics business and co-head of the GEAR lab, recently posted on X criticizing the current state of the robotics industry. Looking back at developments in robotics in the year now coming to an end, he argues that the robotics field remains in a state of chaos, and that its development direction may be wrong.
The biggest recent news in AI chip development is Nvidia's non-exclusive technology licensing agreement with Groq. Nvidia invested US$20 billion to acquire Groq's technology license and onboard its core engineering team.
As frontier AI models reach practical usability, competition is shifting from incremental improvements in model performance to a broader battle over distribution, application integration, and cost-effective deployment, according to DIGITIMES analyst Luke Lin. Developers are now racing to secure access to end users via devices and platforms, marking a new phase in AI competition that is expected to last for the next several years.
As AI adoption accelerates, global and US economies in 2025 have been increasingly driven by technology companies' investments in AI infrastructure. Analysts warn that next year's AI server demand will depend on whether deployed AI can generate revenue, distinguishing "good AI" from "bad AI" with unclear monetization.
Taiwan's aging population and declining birth rates have led to an increasingly pressing labor shortage, a major structural issue uncovered through a National Taiwan University (NTU) study commissioned by the National Development Council (NDC) on how demographic shifts impact policies. Under this backdrop, industry experts see high potential for the robotics sector, but cost reduction and improved intelligence remain key challenges for widespread adoption.
On December 30, Chinese AI company Z.ai launched a share sale to raise HK$4.35 billion (approx. US$560 million), aiming to become the first large language model (LLM) developer listed in Hong Kong amid a tech IPO surge, with a scheduled listing for January 8, 2026. Meanwhile, Z.ai has recently launched its latest open-source model, GLM-4.7, signaling a shift toward enterprise-focused AI and broader global adoption. As the company prepares for a 2026 IPO, the release highlights the growing influence of independent developers in a market increasingly shaped by hyperscale competitors and ecosystem distribution.
Under ongoing geopolitical risks and rising tariff barriers, Foxconn Technology Group's smartphone assembler FIH Mobile plans to shift part of its new product introduction (NPI) operations from China to Vietnam starting in 2026, sources say. This move marks a significant step as the industry adapts to heightened supply chain resilience demands amid US-China tech tensions.
The field of AI robots has now entered a phase of rapid development and iteration, as evidenced by optimistic predictions from Nvidia CEO Jensen Huang and Tesla CEO Elon Musk. As noted by the National Science and Technology Council (NSTC), key technologies for building advanced AI robots in the future include semantic reasoning, task decomposition, causal inference, chain-of-thought, multi-step reasoning, and cross-domain generalization capabilities.
Global 5G penetration is expected to overtake 4G in 2027 to become the dominant network technology, according to the latest Mobility Report from Ericsson. In addition to this pivotal milestone in mobile communications, the report also points to the beginning of the standardization process for 6G, which is expected to reach 180 million users by 2031, with advanced markets such as the US, China, Japan, and South Korea leading the way in terms of adoption.
Taiwan-based tech products distributor Weblink International is riding a surge in memory prices and robust demand for AI servers, positioning the company for strong growth in 2026. President Dave Lin said both segments continue to face supply constraints, but remain among Weblink's most promising growth drivers.
Facing severe memory shortages and rapidly rising prices, global PC brands are implementing strategies to offset cost pressures, including price increases or maintaining prices with downgraded specifications. Channel partners point out that memory accounts for nearly 20% of a PC's total bill of materials (BOM), and recent sharp hikes in memory prices have forced many vendors to quietly adjust their PC pricing.
The rapid expansion of AI data centers is triggering unprecedented electricity demand, placing Taiwan's power grid under growing strain. Taiwan Power Company (Taipower) Chairman Wen-sheng Tseng warns that land scarcity, urban density, and climate risks are converging into a critical infrastructure challenge.