Daily: China's 15th Five-Year Plan; Nvidia x G42; TSMC's latest CoPoS tech
6.5 min read.
Highlights
China’s 15th Five-Year Plan. Top Chinese policymakers are in discussion with business leaders to incorporate the private sector into China’s 15th Five-Year Plan (2026-2030). This includes AI and semiconductors. The Chinese state has always had a fickle relationship with the private sector, so we will have to wait and see if the state makes good on their promise to foster innovation.
Nvidia x G42. Nvidia teams up with G42, UAE’s top AI company, to build data centres powered by Blackwell chips. The Middle East is increasingly taking on a central role in the AI space. Following President Trump’s visit to the Middle East in May, you could also see the Middle East as part of a grand strategy in the framework of U.S.-China competition. I would also put Southeast Asia in that middle ground, though we haven’t seen too much in that region yet. Something to stay tuned for.
TSMC’s latest CoPoS tech. Following TSMC’s CoWoS technology, the Taiwanese company is now developing CoPoS (Chip-on-Panel-on-Substrate). They will test CoPoS technology in 2026 and expects to mass produce in 2029. Nvidia will likely be the first big customer for CoPoS.
Thanks for reading.
1. Policy and Geopolitics
1.1
SCMP (06/11): From AI to chips: China courts private tech firms to help drive next 5-year plan
China has signalled growing support for the private sector, as the head of the government department that oversees economic reform sat down with representatives of the tech industry to gather input ahead of Beijing’s next five-year development blueprint, analysts said.
Zheng Shanjie, chairman of the National Development and Reform Commission (NDRC), met leaders from five private enterprises to “gather opinions and suggestions for the scientific formulation of the ‘15th Five-Year Plan’, with a focus on technological innovation”, according to a statement published by the NDRC on Tuesday.
Zheng said private enterprises could play a crucial role in helping Beijing formulate its next plan, which will cover the years 2026 to 2030. The private sector “possesses strong innovation momentum, great potential and abundant vitality”, he added, making private companies “a key force in developing new quality productive forces”.
1.2
Nikkei (06/11): China's critical mineral curbs shake AI data center suppliers
Tech companies building AI data centers for Nvidia, Amazon and Google are counting on a breakthrough in U.S.-China trade negotiations as their stockpiles of critical minerals and related materials run painfully low.
China tightened its restrictions on critical elements used in tech manufacturing in April amid trade tensions with the U.S. Since then, prices of these elements have surged, and supplies of materials that use them, such as low-temperature solder paste, can be measured in weeks, executives from multiple suppliers told Nikkei Asia.
1.3
FT (06/11): Europe’s AI computing shortage ‘will be resolved’ soon, says Nvidia chief
Nvidia chief Jensen Huang said Europe’s shortfall in computing power for artificial intelligence “will be resolved” soon, as the continent pushes to catch up with the US and China in the race to capitalise on the fast-developing technology.
At least 20 AI data centres are planned across Europe in the next few years, Huang said at the VivaTech conference in Paris on Wednesday, as he predicted a tenfold increase in the region’s data centre capacity within the next two years.
This will include five “gigafactories” — sites carrying hundreds of thousands of Nvidia’s powerful graphics processing units, the chips needed to train and run large language models.
1.4
Reuters (06/11): Qualcomm launches AI R&D centre in Vietnam
U.S. chipmaker Qualcomm has launched an artificial intelligence research and development centre in Vietnam, the company said on Tuesday.
Its researchers will focus on advancing generative and agentic AI solutions across smartphones, personal computers, XR (extended reality), automotive and IoT applications, it said in a statement.
2. Economy, Finance, and Business
2.1
TrendForce (06/11): Micron Rumored to Hike DDR4 Prices by 50% as Samsung Runs Out Early; Industry Says Pricing in Chaos
As Samsung winds down DDR4 production with final orders due in June, module makers seem to be already scrambling for supply. Amid the tight market, rumors say Micron has hiked June DDR4 prices by 50%, sending 8Gb and 16Gb spot prices soaring.
Industry sources indicate that DDR4 pricing is in chaos, with some quotes jumping as much as 100%. Another Etnews report noted that Samsung has raised DRAM prices for the first time in over a year amid tight supply, with DDR4 seeing the sharpest jump.
The report suggested that Samsung, finalized new pricing terms with major clients in early May, has raised DDR4 prices by around 20%.
2.2
Bloomberg (06/11): Nvidia Teams Up With G42 for UAE Data Centers in Mideast AI Push
Nvidia Corp. will collaborate with a G42 unit to build data centers powered by its Blackwell chips, a sign of how the United Arab Emirates’ access to the latest generation of semiconductors has improved since US President Donald Trump’s May trip to the Middle East.
Khazna, a data center developer majority-owned by G42, plans to use the chips to develop artificial intelligence clusters of as much as 250 megawatts, Khazna said in a statement Wednesday. Nvidia certified the design of the facilities to support its Blackwell architecture, it said.
2.3
Bloomberg (06/11): DayOne Gets $3.5 Billion Funding for Johor Data Centers
DayOne Data Centers Singapore Pte. secured 15 billion ringgit ($3.5 billion) of multicurrency financing to support its green data centers in Malaysia’s Johor state, according to a statement from Oversea-Chinese Banking Corporation Ltd.
OCBC and its Malaysian unit OCBC Bank (Malaysia) Bhd. acted as joint coordinators for the syndicated financing, which comprises a 7.5 billion ringgit Islamic financing and a $1.7 billion offshore term-loan facility, the bank said.
2.4
Reuters (06/11): CoreWeave to offer compute capacity in Google's new cloud deal with OpenAI, sources say
CoreWeave has emerged as a winner in Google's newly signed partnership with OpenAI, sources familiar with the matter told Reuters, in the latest example of the voracious appetite for computing resources in the artificial-intelligence industry and the formation of new alliances to meet them.
The so-called neocloud company, which sells cloud computing services built on Nvidia's graphics processing units, is slated to provide computing capacity to Google's cloud unit, and Alphabet's Google will then sell that computing capacity to OpenAI to meet the growing demand for services like ChatGPT, the sources said. Google will also provide some of its own computing resources to OpenAI, added the sources, who requested anonymity to discuss private matters.
3. Technology
3.1
TrendForce (06/11): TSMC Reportedly Gears Up for CoPoS Mass Production by 2029, Tapping NVIDIA as First Client
With big techs led by NVIDIA betting on TSMC’s CoWoS, the foundry giant is moving ahead with its CoPoS (Chip-on-Panel-on-Substrate) technology amid strong AI demand. According to MoneyDJ and the Economic Daily News, TSMC’s first CoPoS pilot line is set for 2026, with mass production targeted by 2029.
MoneyDJ highlights NVIDIA as the likely first big customer for TSMC’s CoPoS, while Economic Daily News points out that CoPoS—built for high-end applications like AI—extends CoWoS-R for Broadcom and CoWoS-L for NVIDIA and AMD.
According to the reports, TSMC’s CoPoS is essentially a square-panel evolution of CoWoS-L and CoWoS-R, swapping the traditional round wafer for a rectangular substrate. Measuring 310x310mm, the rectangular design reportedly offers more usable substrate space than traditional round wafers, boosting output efficiency and cutting costs.
3.2
TrendForce (06/11): Micron Ships HBM4 Samples with 1ß Process to Multiple Customers, Reportedly Including NVIDIA
U.S. memory giant Micron announced on June 10 that it had shipped 36GB 12-high HBM4 samples to key customers—reportedly including NVIDIA and other major tech players, as noted by South Korean media outlet Business Post.
The report suggests that Micron is now the second of the big three memory makers to reach the HBM4 sample stage, following SK hynix, which took the lead by delivering samples to customers on March 19.
According to Micron, it has started shipping 36GB 12-high HBM4 samples to major customers, gearing up for mass production in 2026 to align with next-gen AI platform launches. Tom’s Hardware earlier noted that NVIDIA’s 2026-bound Rubin will drive the shift from HBM3/HBM3e to HBM4, with Rubin Ultra adopting HBM4e.
3.3
Bloomberg (06/11): Nvidia’s Huang Sees Quantum Computing Reaching Inflection Point
Nvidia Corp. Chief Executive Officer Jensen Huang said Wednesday that quantum computing is reaching an inflection point and will be powerful enough in the coming years to help “solve some interesting problems” globally.
Quantum computing, which technology companies have been developing for decades, is set to take off with systems rapidly becoming “more robust, higher performance and more resilient,” Huang said during a keynote speech at a company event being held jointly with the VivaTech 2025 conference in Paris.
3.4
Nikkei (06/11): China's Xpeng unveils SUV with in-house AI chip as Nvidia alternative
China's Xpeng on Wednesday showed off its first vehicle equipped with a proprietary smart-driving chip, a move designed to distance the automaker from U.S. chipmaker Nvidia.
The new G7 SUV comes equipped with Xpeng's Turing chip, named after computer science pioneer Alan Turing and billed as a boost for the company's intelligent cockpit and autonomous driving capabilities.
-