☕️ Good morning friends,

We personally love Taiwan's beef noodles best, but the rest of the world loves the chips from TSMC. That's exactly what a dispute has now erupted over.

More on that in today's TOP BIT.

Also in today's issue:

• Pony AI and Toyota start series production of 1,000 autonomous e-crossovers.

• India's IT model is wobbling: AI agents from Anthropic automate coding and finance.

• Our new section: Chart of the Week. Today featuring the world's top 10 innovation clusters.

Enjoy reading! 📰

👉🏻 Thousands have already watched why Michael moved from Germany to China. Click here if you want to know too!

Asia slides again as tech sell-off deepens: Asian markets extended losses after another tech-led rout on Wall Street. South Korea was hit hardest, with the KOSPI down sharply as Samsung and SK Hynix slid. Japan briefly dipped before stabilising, as risk aversion spread across equities, crypto and commodities.

TOP BIT

Taiwan Says No: US Chip Plan "Impossible"

Taipei draws a clear red line: Washington's plan to relocate 40% of Taiwanese chip production to the US is simply "impossible."

• Despite new billion-dollar deals, Taiwan guards its technological supremacy as geopolitical life insurance.

What the US Demands

Commerce Secretary Howard Lutnick laid out the 40% target in January – as part of the latest trade agreement with Taiwan.

If Taiwan doesn't cooperate, Lutnick threatens tariffs of 100%.

What Taiwan Offers 

TSMC is already investing $165 billion in Arizona factories. But Taipei draws a red line: The most advanced manufacturing stays in Taiwan.

• The so-called "N-2 rule" stipulates that overseas factories may only use technologies at least two generations behind the latest standard.

The "Silicon Shield" 🛡️

Taiwan produces around 90% of all cutting-edge chips (<3 nm). This dominance serves as a strategic shield: The more indispensable Taiwan is to the global tech industry, the greater US interest in its security.

A massive chip exodus would undermine this protection.

⚠️ Reality Check

Semiconductor analysts consider US goals illusory.

• Production costs in the US are 30–50% higher

• acute skilled labor shortage

• and sheer logistical complexity.

A shift of this magnitude is physically barely feasible by the end of Trump's term in 2028.

📊 All details & data: CNBC, Reuters, Yahoo Finance

OUR PARTNER

When training takes a backseat, your AI programs don't stand a chance.

One of the biggest reasons AI adoption stalls is because teams aren’t properly trained. This AI Training Checklist from You.com highlights common pitfalls and guides you to build a capable, confident team that can make the most out of your AI investment. Set your AI initiatives on the right track.

Get the checklist.

CHART OF THE WEEK

The two most innovative city clusters in the world, Shenzhen-Hong Kong-Guangzhou and Tokyo-Yokohama, filed 117,542 PCT patents in 5 years.

That represents 20% of all global applications.

🏅 China accounts for 24 of the top 100 clusters, the US 20, Germany 8.

MARKET BIT

China industrializes robotaxis: Toyota and Pony.ai take autonomous cars to scale

China’s robotaxi pioneer Pony AI and Toyota have kicked off commercial production of their jointly developed autonomous vehicles. The first of 1,000 all-electric Toyota bZ4X crossovers rolled off the line this week at the Guangzhou Toyota plant.

🤝 The joint venture between Toyota and Guangzhou Automobile Group is manufacturing vehicles equipped with Pony AI’s latest autonomous driving software.

Goal: Expand Pony AI’s robotaxi fleet to more than 3,000 vehicles by year-end.

The details

Toyota is tapping China’s robotaxi boom to scale its EV platform. Pony AI, in turn, gains proven automotive-grade manufacturing quality for its fleet rollout across major Chinese cities.

• Robotaxis in China typically operate at Level 4 autonomy: no human intervention in most situations—within clearly defined operational domains (geofenced areas).

Complicated name, simple logic: the bZ4X is one of three vehicle models Pony AI is deploying in parallel across multiple Chinese markets.

Human in the loop: Pony AI now needs roughly one support operator per ~30 robotaxis to handle edge cases.

Market reality

At scale, robotaxis are decided not just by software—but by cost curves and fleet operations. China holds two key advantages:

• Regulatory “controlled rollout” in top-tier cities (Beijing, Shanghai, etc.), paired with rapid infrastructure iteration.

• Market size potential: HSBC estimates robotaxis could capture ~6% of China’s taxi/ride-hailing market—an addressable ~US$40bn per year.

Chinese robotaxi players are expanding alongside U.S. rivals such as Waymo, Zoox and Tesla, both domestically and abroad.

The difference: China combines aggressive scaling with established OEM partnerships—unlocking immediate production capacity.

👉 Full story: BusinessTimes, SCMP, Engadget

HEAD OF THE WEEK

🇭🇰 Pansy Ho

♦️From casino heiress to experience architect

Born into one of Asia’s most powerful gambling dynasties, Pansy Ho could have stayed comfortably on the casino floor. Instead, she chose to redesign the entire destination.

As Co-Chair and Managing Director of MGM China, Ho has been at the forefront of Macau’s biggest transformation in decades. With Beijing urging the city to move beyond gambling, she shifted the focus toward entertainment, culture, and high-end experiences.

Her strategy is clear: make Macau a destination again, not just a betting hub.

HIGHLIGHTS 

🇮🇳 AI agents put pressure on India’s IT outsourcing model: New tools from Anthropic show how AI can automate coding, legal, finance and sales work, directly challenging the traditional hourly billing model of Indian IT service providers. Clients are asking less about headcount and more about outcomes. The industry is not being displaced, but redefined: away from pure manpower and toward scaling, integration and operation of AI-powered software.

🇨🇳 Chinese machine vision outpaces human reaction: Researchers from China, the UK and the US have developed a safety system for autonomous vehicles that detects hazards faster than the human brain. While humans need around 0.15 seconds to react, machines previously took about half a second. At 80 km/h, that difference translates into roughly 13 meters of additional braking distance.

🇨🇳 ByteDance and Alibaba take aim at Google’s Nano Banana: TikTok owner ByteDance unveiled its Seedream 5.0 image model, positioning it as a cheaper alternative to Google’s popular Nano Banana editor. At the same time, Alibaba Cloud released Qwen-Image-2.0, which it says delivers better results for Chinese characters than Google’s Nano Banana Pro. Both models are already available in beta across multiple platforms.

🇰🇷 South Korea expands medical school intake: The government plans to increase annual medical school admissions by an average of 668 seats through 2031, with most of the growth allocated to 32 regional medical schools outside Seoul. The move follows backlash against a previous plan to add 2,000 seats, which triggered an 18-month doctors’ strike. South Korea has about 2.7 doctors per 1,000 people, well below levels in other developed countries.

FORTUNE COOKIE

Morning coffee for the team! ☕

MORE ASIABITS

• The Asiabits Podcast on YouTube. 

• The Asiabits Podcast on Spotify. 

YOUR FEEDBACK