Know Asia's
Business changes,before it hits your market.

The legal foundation of Donald Trump's global trade war has developed cracks. In a historic ruling, the US Supreme Court declared the global tariffs imposed by the Trump administration null and void.

For Beijing, this victory comes at just the right time – only weeks before the crucial summit meeting between Trump and Xi Jinping.

What the court decided

Trump had misused the International Emergency Economic Powers Act (IEEPA) to impose blanket tariffs. The judges completely overturned the legal basis.

Trump's reaction: Within hours, he imposed 15% global tariffs under a different law (Section 122 Trade Act) – which, however, requires congressional approval after 150 days.

The winners and losers of the reordering

Timing couldn't be better – for Beijing

Trump is traveling to Beijing at the end of March. Through the ruling, he has lost one of his most important "threats." China, on the other hand, is appearing more confident than ever before:

• Record surplus: With a trade surplus of $1.2 trillion last year, Beijing has proven that it can successfully redirect exports even without the US market.

• Raw materials leverage: China controls global supply chains for rare earths. Restrictions on magnets and minerals hit the US defense industry (e.g., F-35 jets) hard.

• Counter-demands: It is expected that Xi will demand massive concessions on the Taiwan question and US technology export controls in exchange for agricultural purchases (soybeans).

Shenzhen-based startup AI² Robotics has raised more than 1 billion yuan (~$144 million) in a Series B round, bringing its valuation to just under $1.5 billion.

The investor lineup includes Baidu, CRRC Corporation Ltd., several companies from the Tesla ecosystem, and Guotai Haitong Securities.

• The signal: with a major securities firm on board, this is not just growth capital — it’s IPO preparation.

From demo hype to capital markets readiness

Dancing robots are nice, but China’s embodied AI sector is clearly shifting from prototype showcases to industrialization. What matters now are scalable, repeatable deployments in real production environments.

AI² Robotics was founded in 2023 by Guo Yandong, a former chief scientist at XPeng and a Purdue PhD.

🤖 Robots in action:

• The wheeled dual-arm robot AlphaBot 2 is deployed in automotive and semiconductor manufacturing for sorting and labeling tasks.

• Systems are trained using a combination of simulation and real-world data.

• The company has also open-sourced its own visual-language-action (VLA) models.

The broader backdrop: in 2025, more than 20 Chinese robotics startups each raised around 1 billion yuan. Companies such as Galaxea AI, Galbot, Leju Robotics, and Unitree are preparing A-share listings or eyeing Hong Kong IPOs.

Listing or liquidation?

A year ago, a $42 million round was the largest financing in the sector. Today, five companies have each raised over $140 million within eight weeks.

The size of the round isn’t the key factor. The timing is. Capital is concentrating. The competition is shifting from “who has the best model?” to “who can generate recurring industrial revenue?”

The playbook is clear: reach the public markets quickly before consolidation tightens access to growth capital.

• Public listings are becoming a financial safety net. Investors point to UBTech in Hong Kong as an example, despite ongoing losses, its listing status provides strategic stability and continued access to capital.

But: industrialization requires a different level of capital than building prototypes. 2026 will be the inflection point. The companies that can convert embodied AI demos into durable industrial revenue will stay in the game.

The EU has excluded Chinese organizations from critical areas of its flagship research program Horizon Europe starting in 2026.

The measure's goal: protecting intellectual property and securing Europe's technological sovereignty.

What gets excluded

The €93 billion program remains blocked for China in the following areas:

Reactions from China

"The policy will not harm China much and could make Europe appear more isolated."

Other experts warn: China is now the world's leading producer of AI research. The restrictions could long-term reduce EU access to Chinese talent and contribute to a more fragmented global research landscape.

Hyundai Motor Group will invest 10 trillion won (around $6.9 billion) over five years in Saemangeum, North Jeolla Province.

The goal: a future-focused cluster for AI, hydrogen and robotics on a 409-square-kilometer reclaimed land project, roughly two-thirds the size of Seoul.

The strategic context

The 10 trillion won commitment is part of Hyundai’s broader 125 trillion won domestic investment plan through 2030. Saemangeum is the first major concrete project under that roadmap.

• Nvidia backbone: Hyundai is doubling down on AI infrastructure. Chairman Chung Eui-sun has already agreed with Jensen Huang to deploy 50,000 Blackwell GPUs.

That computing power is intended to accelerate autonomous vehicles, robotics and digital twins in manufacturing.

Hydrogen strategy: Hydrogen has been a core pillar of Hyundai’s long-term strategy for years. Solar-powered electrolysis would enable the production of “green” hydrogen.

Robotics is also central to the plan: from the humanoid Atlas to industrial wearables and mobile platforms. Saemangeum is set to become a testing ground for smart mobility and autonomous systems.

Why Saemangeum?

Ample land, strong solar conditions and political backing: Hyundai plans to build an AI data center, large-scale green hydrogen electrolysis facilities and robot manufacturing operations there. Local solar installations are also under consideration to generate and consume energy on site.

Beyond land and sunlight, aggressive investment incentives are a draw:

• Long-term land leases at symbolic rates

• Employment subsidies

• Government support and coordination

Seoul is pushing to industrialize regions outside traditional hubs such as Ulsan — and Saemangeum is emerging as a flagship project.

An Indian university was kicked out of the prestigious AI Impact Summit in New Delhi on Wednesday. A professor had passed off a commercially available robot dog from China as her own groundbreaking innovation.

India actually wanted to use the summit to present itself as a new AI superpower alongside the US and China. But then this happened:

• The fraud: A professor from Galgotias University proudly explained on state television that the robot dog "Orion" had been developed at their own center of excellence.

• The exposure: Internet users identified the machine as a Unitree Go2 within minutes.

• The consequence: The university had to clear its booth, power was cut, and India's IT minister hastily deleted his celebratory posts on X.

High-stakes summit becomes chaos stage

The "robo-lie" is particularly explosive because the summit was attended by the absolute world elite: Prime Minister Modi, Emmanuel Macron, OpenAI CEO Sam Altman, and Google CEO Sundar Pichai were on the guest list.

India is investing massively in "IndiaAI" to develop sovereign AI models and not be dependent on US tech giants.

The country wants to establish a "third way" between the US and China – with open-source models, 7,000 public datasets, and planned investments of $200 billion in data centers.

But the path is still rocky so far.

Fei-Fei Li, known in the industry as the “Godmother of AI,” has raised $1 billion with her startup World Labs.

🤝 Rocket start: Li only founded World Labs in 2024. In September 2024, World Labs had already raised $230 million. Since then, the valuation has more than doubled. The new round values the company at around $5 billion.

The details

World Labs is developing “Spatial Intelligence”: AI that can understand and model the physical world in 3D. The first product is called Marble and generates interactive 3D worlds from images or text.

The round is led by Autodesk with $200 million alone, joined by Andreessen Horowitz, Nvidia, AMD, Fidelity, and Singapore’s Sea Limited.

Who is Fei-Fei Li?

Born in Beijing and later moving to the United States, she is now a central voice in the U.S. AI scene.

She is one of the most influential researchers in modern AI, especially due to ImageNet, the dataset that massively accelerated computer vision in the 2010s.

Li previously served as Chief Scientist for AI/ML at Google Cloud before returning to Stanford and later launching World Labs.

Why everyone is betting on Spatial Intelligence

The thesis: AI must understand the physical world in 3D, not just process text and images. This opens up applications in robotics, AR/VR, and scientific simulation.

Meta’s chief scientist Yann LeCun is pursuing a similar approach with AMI Labs, and Google DeepMind is developing its own system called Genie. The participation of Singapore’s Sea Limited highlights the interest of Asian tech giants.

The investor logic: After the LLM boom, spatial intelligence will be the next growth market.

Japan and the US have agreed on the first three megaprojects of their $550 billion investment package. Trump celebrates it as proof of his tariff policy. Tokyo is under pressure – and pays.

"The scale of these projects are so large, and could not be done without one very special word, TARIFFS."

The three projects

The gas power plant alone equals nine nuclear reactors and could supply 7.4 million households. Japanese companies like Toshiba, Hitachi, and Mitsubishi Electric are to deliver turbines.

The fine print

The profit participation is extremely one-sided. After repayment of the initial investment, the profit split shifts to 90:10 in favor of the US, according to reports.

• Additionally, after Trump's approval, Tokyo has only 45 days to physically transfer the promised funds.

⚠️ Reality check

"Strategic, but hardly bankable," is how Japanese financial circles describe the projects. Politicians and corporate leaders have openly criticized Trump's tactics as "blackmail."

But dependence on the US market and especially Washington's security policy role against China, North Korea, and Russia left no choice.

A Blackstone-led consortium is putting up to $600 million in equity into Neysa, an Indian startup founded in 2023 that builds and operates GPU-based AI infrastructure for enterprises.

• On top of that, Neysa plans to secure another $600 million in debt financing, bringing the total to $1.2 billion for a company that isn't even three years
  old.

🤝 Blackstone's Amit Dixit, head of Asia private equity, will partner with Neysa co-founder and CEO Sharad Sanghi to scale the business.

The details

Neysa deploys GPU clusters inside India - for banks, tech firms, hospitals and government agencies. The fresh capital will fund the rollout of more than 20,000 GPUs across the country, specifically for AI training and high-performance computing.

Joining Blackstone in the round are Teachers' Venture Growth, TVS Capital, 360 ONE and Nexus.

The key thing here: Neysa designs and operates its systems entirely within India. Data stays onshore. Modi's government is pushing hard on "sovereign compute" under the IndiaAI Mission banner.

The bigger picture

For Blackstone, Neysa fills the India gap in a global AI infrastructure portfolio that already includes QTS and CoreWeave in the US, AirTrunk across APAC and Firmus in Australia.

The timing could hardly have been better. On the same day, Adani announced $100 billion for AI data centers by 2035. Modi had flown in Nvidia CEO Jensen Huang, Google's Sundar Pichai and half a dozen other tech chiefs for the India AI Summit in New Delhi. AMD simultaneously signed an expanded AI partnership with Tata.

India is clearly serious about this. Whether the country can actually build compute capacity as fast as the cheques are flying is the real question.

Radical shift in the world's most technologically advanced car market: China is ending the era of buttonless cockpits.

China's Ministry of Industry has published a draft of new safety regulations mandating physical buttons for life-critical functions.

From July 1, 2027, all newly manufactured vehicles in the People's Republic must have "real" controls again.

The new mandatory buttons

The following functions may no longer be operated exclusively via touchscreens:

• Gear shift and turn signals

• Hazard lights and window controls

• Emergency call (eCall)

• Minimum size: 10 x 10 mm per button

The reason: distraction at the wheel. In many electric cars, even mechanical functions run through the central display. In the Tesla Model Y, the driver has to tap the display to change gears.

Not the first reversal

China is systematically cleaning up design trends that are problematic from a safety perspective:

Retractable door handles can jam after accidents and trap occupants. Yoke steering wheels don't meet airbag standards – according to the ministry, the steering column is involved in 46% of all driver injuries.

China is not alone. The European testing organization Euro NCAP has also already announced that from 2026, it will no longer award five stars if safety-critical functions can only be operated via screens.

• The message is clear: futuristic aesthetics are all well and good – but not at the expense of safety.

MiniMax surged as much as 30% in Hong Kong before closing up 25%. Since its IPO in January 2026, the stock is now up more than 400%.

Rival Zhipu also posted double-digit intraday gains. Both names are currently among the hottest AI bets in Asia..

The driver

Optimism around China’s generative AI startups is rising, fueled by model upgrades ahead of Lunar New Year, a period when platforms traditionally compete for maximum user traffic.

• MiniMax last week released version M2.5 of its flagship model.

On the SWE benchmark, M2.5 performs “very close” to Claude Opus 4.6, with analysts pointing to “significant performance improvements.”

Why so hot?

First: China is investable again.

• After years of regulatory uncertainty and weak tech performance, investors are searching for a new growth anchor. Chinese AI offers exactly that: structural growth, strategic relevance and state backing.

Second: valuation arbitrage.

• U.S. AI stocks trade at extreme multiples. In Hong Kong, newly listed AI players are entering the market at far lower starting valuations.

Third: technological catch-up.

• MiniMax’s M2.5 and Zhipu’s GLM models are being publicly benchmarked against U.S. frontier models. “Close enough” is sufficient to support the thesis that China is not permanently behind in the LLM race.

Fourth: geopolitical logic.

• Semiconductors, AI and compute capacity are strategic sectors. Investors understand that these areas are politically prioritized in China, reducing the risk of sudden domestic policy headwinds.