スプーキーSpookyブーギーBoogie

Last week “Secretary of War” Pete Hegseth insulted Americans by claiming that a 50 percent increase in the US military budget – from an incomprehensible one trillion dollars to an impossible one and a half trillion – was a “fiscally responsible investment.”

“Thanks to President Trump’s $1.5 trillion defense budget, this War Department has moved from bureaucracy to business,” he said last Thursday.

In a way he was right, though. The huge increase is much more about “business” than what is needed to protect the United States from potential invasion.

But it isn’t the kind of “business” that most supporters of free markets would applaud. On the contrary, this is the business of transferring massive amounts of wealth from the struggling middle and working classes to the well-connected Beltway elite based on lies and scare tactics.

The US mainstream media is crucial in manufacturing the fairy tale that if we don’t mortgage our children’s and grandchildren’s future to finance this obscene military budget, we will be attacked or invaded by some evil foreign power.

It’s not difficult to do a little research and see why the mainstream – and even some “independent” – media outlets push these scare tactics: they are owned or funded by giant corporations with close ties to military contractors.

This unhealthy relationship is known as “corporatism” – the intermingling of pseudo-private companies with the government. It is the precursor to actual fascism, where the government takes a stake in such companies. We’re getting there faster than most Americans understand.

The whole scam is not about protecting the citizens of the United States. It’s about protecting the US empire overseas, which actually harms the citizens of the United States.

Yes, they rob us to fund their empire and lie to us that it keeps us safe. Nothing could be further from the truth. Our constant military interventions on virtually every continent of the globe only build resentment among the rest of the world’s population. Anyone who thinks people overseas welcome US bombs has been watching too much Fox News or reading too much Washington Post.

And what do we get for the most expensive military on earth – larger than the combined militaries of the next dozen or so countries? Not much. Iran’s military budget is less than one percent of ours, yet Iran destroyed or disabled every US military base in the Middle East.

It turns out that Iran has destroyed dozens of multi-million dollar US spy drones – and several near-billion dollar spy radar stations – with their own drones costing mere thousands of dollars each.

The US surprise attack was supposed to make Iran cower and beg for mercy, but it did the opposite: it showed that despite the trillions extorted from Americans for the most expensive military on earth, the US military can no longer win the wars that US presidents illegally force them into fighting.

The US military continues to fight World War II – with massively expensive aircraft carriers that do not dare get close to combat – while warfighting has evolved into something entirely different.

The only good thing about the Iran war is that it demonstrates how much the special interests have lied to us about the need to continue our suicidal military spending increases.

It was never about protecting the United States. It is about protecting the ever-growing bank accounts of the special interests at the expense of the rest of us. It needs to stop. Now.

Li Hongzhi, an award-winning former head of GenAI at Microsoft Asia, has joined Tongji University, one of China’s leading universities.

Li started his first job at technology giant Microsoft immediately after obtaining his PhD from Columbia University.

For more than 10 years, he worked at Microsoft Research – the company’s subsidiary responsible for basic and applied research in computer science, software engineering and hardware design.

Li was the head of Microsoft AI Asia’s GenAI Group before joining Tongji.

According to the university, Li recently returned to China and is a distinguished tenured professor at its Institute of AI for Engineering.

The South China Morning Post has contacted Li for comment.

He earned his PhD in computer science in 2016. Before that, he received his bachelor’s and master’s degrees in computer science from Zhejiang University in 2010 and Columbia in 2012.

While at Columbia, Li won the Grand Challenge at ACM Multimedia 2012, a global conference on multimedia research.

The ACM Multimedia Grand Challenge highlights practical, real-world multimedia problems posed by industry and academia, challenging researchers to benchmark their AI, streaming and vision algorithms on common data sets and platforms.

After obtaining his PhD, Li joined Microsoft Research headquarters in Washington state. Over the 10-plus years he worked there, he gradually rose through the ranks.

He previously held positions as principal researcher and principal architect, as well as principal applied science manager.

His research interests focused on machine intelligence, including multimodal content analysis and cloud computing.

Li co-authored the influential paper “Rethinking Classification and Localisation for Object Detection”, which has proved foundational in the field of computer vision and been cited more than 1,000 times.

Published in 2020, the paper proposed the “double-head method” to solve the distinct requirements of classification and localisation in artificial intelligence.

The method refers to a strategy of using two specialised “heads” or mechanisms to handle different parts of a complex task simultaneously.

At Tongji, Li’s boss is Hua Xiansheng, executive dean of the university’s Institute of AI for Engineering.

Hua worked at Microsoft Research as well for more than 14 years.

According to its official website, the Institute of AI for Engineering at Tongji is a newly established top AI research institution in Shanghai.

Its core concept focuses on “AI4E (AI for engineering)” and it aims to make breakthroughs in important technologies for foundational models and intelligent agents in engineering.

A year ago, many questioned whether Chinese AI chips could become more than emergency stopgaps for restricted US tech. Now, Beijing is building a commercially viable alternative AI stack that is less dependent on American goodwill.

That transition became clearer last month when DeepSeek released V4, a 1.6-trillion-parameter model that it said outperformed OpenAI’s GPT-5.2 in several coding benchmarks. Crucially, the technical report confirmed the AI model had been formally optimised on Huawei’s Ascend 950, a next-generation neural processing unit (NPU) designed for large-scale AI model training.

For the first time, one of DeepSeek’s top-tier AI models was no longer solely dependent on Nvidia hardware and, in some areas, ran more efficiently on Chinese chips.

“Jensen’s concern has become a reality,” said Lian Jye Su, chief analyst at research firm Omdia, citing DeepSeek V4’s adaptation to both Huawei chips and Huawei’s flagship Compute Architecture for Neural Networks (CANN) – widely viewed as China’s closest equivalent to Nvidia’s dominant CUDA software ecosystem.

Within 24 hours of DeepSeek V4’s launch, Huawei was joined by seven domestic chip vendors – including Cambricon Technologies, Moore Threads, Hygon Information Technology and Biren Technology – which announced they had fully adapted to the model.

The developments reflect a broader transformation in China’s AI strategy. Instead of trying to rival the United States solely in cutting-edge chips, domestic firms are increasingly focusing on inference – that is, running and deploying AI models at scale.

“We are already seeing AI chipsets from Huawei, Cambricon, Moore Threads and MetaX gaining momentum in China. Alibaba Cloud’s PPU and Baidu’s Kunlunxin are also being deployed for AI training and inference workloads at hyperscale,” said Su of Omdia.

Morgan Stanley analysts underscored this trend in a recent report, which argued China’s AI graphics processing unit (GPU) market had shifted “from whether domestic chips can participate” to “which vendors will win meaningful share as inference demand scales”.

“China is narrowing the US lead in AI compute not simply at the chip level but also through system-level innovation, supply-chain localisation and increasingly attractive inference economics,” wrote the analysts led by Charlie Chan, a technology researcher leading the investment bank’s Greater China semiconductor coverage.

Zhang Yunquan, a professor specialising in high-performance computing at the Chinese Academy of Sciences, said rising demand for computing power had made inference-friendly chips an area where China could compete internationally. But he cautioned that China remained behind in high-end training chips – with Huawei leading domestic efforts to close the gap.

“Over three to five years, with continued iteration on chipset yields and CANN maturity at cluster scale, domestic chips could handle most of the domestic training workloads,” said Su of Omdia.

But chips are only half the story, as Nvidia’s greatest strength has never been just computing power. The dominance of CUDA – its proprietary software ecosystem and programming model underpinning AI development – has made it extremely difficult for users to migrate to other systems.

DeepSeek V4’s shift towards Huawei’s CANN, however, suggests China may be starting to loosen Nvidia’s stranglehold.

“As the AI chipset ecosystem in China continues to mature, we will see … more Chinese model providers build AI models optimised for Chinese AI chipsets from day zero,” Su said.

The emergence of a viable domestic AI stack in China is also shifting attention to market share and commercial scale. Morgan Stanley analysts estimated that China’s AI accelerator market could reach US$67 billion by 2030, with the domestic self-sufficiency rate surging from 33 per cent in 2024 to 86 per cent.

Beyond Huawei, the investment bank also highlighted Cambricon, Iluvatar Corex and MetaX as the companies most likely to benefit from the chip localisation strategy.

According to the report, Huawei’s Ascend 950 and Cambricon’s Siyuan 690 can outperform Nvidia’s H20 by 50 to 150 per cent in different scenarios when measured by tokens per second – a crucial metric in China’s inference-driven market that reflects both hardware performance and software optimisation.

Most of the leading domestic chips also carry lower costs, with market prices for Huawei’s Ascend 950PR, MetaX’s C600 and Moore Threads’ MTT S5000 cheaper than both the H20 and H200, according to the report.

The authors also identified a fundamental shift in China’s AI chip procurement – from “policy-driven” to “profit-driven” – as major AI companies rush to monetise amid rising token consumption. That means firms able to deliver ready-to-use accelerators quickly could be best positioned to see surging revenue growth.

But as China’s domestic AI ecosystem advances, uncertainty is growing over the future trajectory of US export controls – and the wider technological rivalry between Washington and Beijing.

Lutnick, the US commerce secretary, said in April that Trump had struck a “delicate balance” on technology sales, given his cordial personal relationship with Xi.

That balancing act is expected to feature at the coming summit, though the outcome of the talks could be complicated by a stark divide within Washington over the scope and scale of tech restrictions.

According to Marro at the EIU, the executive branch had shown some willingness to keep China dependent on US technology, as seen in the H200 policy, while a hardline Congress was pushing aggressive legislation such as the Match Act.

“This is an area of bipartisan consensus,” he said. “And that’s going to limit Trump’s ability to influence that piece of legislation going forward.”

And yet with both leaders wanting to declare victory at the summit, “marginal progress” was possible, he added. This was echoed by Olson of the ISEAS – Yusof Ishak Institute, who said the talks could see limited gains, such as relaxing restrictions on tech and critical minerals and “some commitments by China on purchases of US products”.

On chips, Eurasia’s Wang argued that Trump was still unlikely to ease controls on the most advanced chips below the 7-nanometre threshold, though export volumes of less advanced chips were likely to rise – suggesting Washington would continue to manage, rather than sever, China’s access to US technology.

US President Donald Trump’s landmark visit to China comes as the US-Iran war disrupts global energy supplies, fuels economic uncertainty and adds fresh strain to Washington-Beijing ties. In this story, part of a series examining how rivalry, interdependence and geopolitical crises are reshaping the relationship between the two powers, we examine how artificial intelligence (AI), chip controls and competing technology ecosystems are redefining US-China rivalry.

Nvidia CEO Jensen Huang was the undisputed centre of attention at a 1,000-guest banquet in the heart of Beijing in July last year, attended by Chinese government officials, diplomats, businesspeople and industrial leaders.

Barely managing a few bites of his dinner, he breezed through a marathon of media interviews and accepted a stream of selfie requests from guests. The rock star treatment even spilled into the hotel lobby when he left, where Huang patiently signed autographs for star-struck fans.

The excitement was clear and perhaps understandable: Huang had arrived in Beijing with news that Nvidia’s H20 chip had just been cleared for export to the Chinese market. A watered-down version of Nvidia’s cutting-edge AI chips, the H20 still outperformed many Chinese rivals and, with tech giants in the world’s second-largest economy locked into Nvidia’s ecosystem, it appeared Washington had all but secured the company’s long-held dominance.

But after Huang flew home, the momentum soon shifted. Beijing started an investigation into the H20 citing security concerns, leading to a de facto import ban, and kept the door shut months later when Washington permitted exports of the more advanced H200, one of the company’s most powerful accelerators for advanced AI models – though still not its strongest.

Then in April, DeepSeek, the company behind China’s most famous home-grown AI model, announced a pivot to using Huawei Technologies’ chips. It was the exact scenario Huang had warned of just six days earlier, calling it “a horrible outcome” for the US.

The developments underscore the rapid rise of China’s AI industry and its decreasing reliance on Nvidia’s technology, even as chips remain at the heart of the US-China rivalry.

And now that US President Donald Trump is heading to China – the first visit to the country by a US president in nearly nine years – all eyes are on whether Huang will be part of the delegation. Either way, the question remains: will he finally manage to get what he wants?

It is no secret that China and the United States are locked in a high-stakes stand-off over the future of AI – and Washington’s export controls on advanced chips are a crucial source of that friction. But analysts cautioned that the looming summit between President Xi Jinping and Trump was unlikely to produce any sweeping breakthrough.

Chips – along with Taiwan, the Iran war and other contentious issues – “will be managed, not resolved” at the meeting, said Han Shen Lin, China managing director for The Asia Group.

That view was widely shared by analysts, with some arguing that export controls had become embedded in the broader strategic rivalry, making them increasingly difficult to roll back.

Ker Gibbs, former president of the American Chamber of Commerce in Shanghai, said technology controls, semiconductors and AI were structural issues that were “too hard to negotiate and too deeply embedded in each country’s strategic priorities”.

Kent Kedl, managing partner at Blue Ocean Advisors, and Stephen Olson, a visiting senior fellow at the ISEAS – Yusof Ishak Institute, also underscored the likelihood of only a limited outcome to negotiations, expecting something closer to a “managed stabilisation”.

Complicating efforts to achieve a serious breakthrough, Washington has in recent weeks sought to strengthen its negotiating hand by increasing pressure on China’s tech sector. The proposed Match Act, which advanced in Congress last month, could force allied nations to mirror US semiconductor export curbs, while the Federal Communications ⁠Commission pushed a proposal to block Chinese labs from testing electronic devices for use in the US.

Laila Khawaja, research director at Gavekal Technologies, said there was little political appetite in Washington to roll back its export controls. If implemented in full or in part, the Match Act could deter China’s progress in advanced semiconductor manufacturing and substantially disrupt current capabilities, she warned, adding that Beijing was likely to use the summit to remind Washington that escalation carried costs.

Against this backdrop, some analysts believe Beijing’s goal may be less about rolling back existing restrictions than preventing further curbs. Wang Dan, China director at Eurasia Group, said the latter would be a great deliverable for Beijing. “What’s even more important is that the existing export controls aren’t escalated,” Wang said.

One of the most closely watched issues during the summit will be whether the two sides can finally reach an understanding on Nvidia’s H200, which could serve as a measure of how far both sides are willing to go in managing technological competition.

In December, Trump announced the AI chip could be sold to “approved customers”, provided that a 25 per cent revenue surcharge flowed to the US Treasury. But US Commerce Secretary Howard Lutnick recently told the Senate that China had still not made any purchases, despite Nvidia’s Huang previously stating that customers had placed orders.

“China is about 40 per cent of the world’s technology industry. To concede that market for the US technology industry is a disservice to our country,” Huang also warned last month, after repeatedly urging the US government to approve sales of the chip to China last year.

Analysts said the drawn-out regulatory process surrounding the H200 illustrates how the chip has become a bargaining chip for broader concessions.

Nick Marro, principal economist for Asia at the Economist Intelligence Unit (EIU), said Beijing might continue to resist purchases of the H200 even after the Xi-Trump summit. The current stand-off, he added, had given it “a golden opportunity … to champion its own domestic supplies”, particularly as local firms gradually closed the gap with overseas rivals.

But according to some analysts, Beijing’s long-term push for semiconductor self-sufficiency was at odds with the immediate computing demands of its rapidly growing AI sector.

Khawaja of Gavekal said Beijing was forced to balance these two priorities and would eventually allow H200 imports, given rising demand and limited advanced chip capacity at home, although “how it frames the decision will depend on the summit”.

“A positive outcome could see it presented as part of an open, cooperative stance; otherwise, approvals may be granted quietly to meet practical demand without signalling concession,” she added.

Wang of Eurasia Group agreed, saying “AI is just too important” to China’s latest five-year plan. “As the semiconductor, AI and robotics industries expand rapidly, there is a massive need to boost computing power,” she said.

Wow, that seems like a pretty random assortment of things! I’ve been interrogating every NPC and Sheikah Stone I come across with the Mask of Truth and haven’t seen anything about that, but I’ll keep my eyes peeled. I’m definitely after the truth thing that’s in the Kakariko well (alluded to by one of the old men taking refuge from Hyrule Castle Town), but I haven’t managed to get at it just yet. In the small bit that I’ve played since my last comment, I got the Golden Scale from the fishing hole, but to my surprise, it was still juuuuust short of letting me get to the bottom of the well and enter the opening (although I did get a surprise Piece of Heart from the lab in Lake Hylia as a consolation prize). And as Adult Link, the well is dry but appears sealed off. I haven’t tried that hard to get in, though, so for the time being I’m not asking for any hints, especially since it might just be something that changes as I progress through the dungeons.

…actually, speaking of which: is the Song of Storms something that will just come in due time? Judging by its placement in the pause menu, it does come after the Song of Time chronologically, but I’m a bit mystified at what I need to do to get it. As Young Link, the guy mentions that he’s trying to write a song about spinning, but neither my spin attack nor my boomerang seemed to entice him. Full disclosure: I accidentally spoiled myself when looking something else up (which is why I really try to avoid doing web searches for things) which indicated that the Song of Storms may be the key to opening those secret holes, but I guess I’ll have to wait and see.

[a few days later] Aha! The solution ended up being pretty simple, but I really enjoy how with this kind of game it’s not like, “Get quest, go to marker A, talk to person X, go to marker B, talk to person Y, complete quest.” Instead, I remembered that there was a guy in Kakariko who mentioned that the windmill was used to draw water from the well, and then when I talked to the guy in the windmill he talked about how he was annoyed by some kid who played a song. The mental block I had was that I expected to learn the song as a kid, but instead I had to learn it as an adult and go back in time! First puzzle that I had to solve with time travel, but I suspect it won’t be the last.

As for the Lens of Truth—I wasn’t expecting a whole dungeon! I thoroughly enjoyed being trolled by all the invisible walls and floors. Leans pretty heavily into horror, what with the piles of flesh and the torture implements—definitely seems like the kind of thing that would stick with you if you played it as a kid back when it first came out! It was my first encounter with a Like Like, which I remembered from its Melee trophy eats shields (put together with what at least one person mentioned, specifically Hylian shields), so I kept it stunlocked with my boomerang and escaped unscathed. Also did a cheeky little skip past a pit by jump slashing across the corner, which made me feel smart.

In the end, it is the Song of Storms that opens those secret holes! It worked on every single one of the non-bombable ones that I’d written down. Entirely possible that there’s a third type (or even more) in OoT, but that’s what I’ve found so far. Surely must be somewhere in the game that hints at this…I know that NES-era games were more likely to require you to just throw stuff at the wall and see what sticks, but in this game it seems like everything is pretty carefully laid out.

Also, is it just me, or is the timing with the frog minigame ridiculously tight? I consider myself to be pretty good at rhythm games, but it took me about two dozen tries to get the fixed song that they start out with and I found the changing ones after that impossible. Wonder if it’s a bug in the original code—I notice that there’s a tweak in Ship of Harkinian to give you unlimited time to complete it (because of course there is!).

Oh, and regarding the Lens of Truth: can you give a general idea of how frequently it’ll actually come in handy? Knowing me, I’ll probably end up scouring every last inch of Hyrule with it regardless, but it’d be nice to know how often I should be hitting paydirt.

If anything, Sheik loves the poetic so I imagine that Japanese would be hard to translate.

Yeah, I was definitely getting lost with her dialogue as well as the sages, which is kinda problematic when that’s where most of the core story is.

There is a way to fix your sword, but I’ll let you find that one out on your own

200 Rupees?! Well, I act indignant, but I’m basically always maxxed out even with the Giant’s Wallet, and if I were ever strapped for cash a single three-bottle run to the Ice Cavern for some Blue Fire to sell to that person who wants to buy stuff in bottles would basically fill it up in one fell swoop. So I’ll just take it as Link being a Job Creator™. One thing that I was expecting to need to spend my money on was the tunics, but I got both of them for free, so there are no other big ticket items to spend on (not that I’ve seen yet, anyway). I even delayed the last part of the mask trading quest because I didn’t want to waste the money, but in the end I sold it for 0 rupees

(also, I see that Biggoron has appeared at Death Mountain’s summit, but it looks like I need to advance further before he’ll forge anything for me. The comment about “Hylian carpenters praise me for my skills” makes me wonder if I need to chat with the carpenter from Kakariko, who I spotted across the broken bridge in Gerudo Valley (another nut which I’ve yet to crack, but all in due time, I’m sure))

riding around with her felt like the world was big in a way previous games couldn’t capture

I do really want to go back and play some of the 2D Zeldas. I bounced off of the original game when I tried it a little while back (might have to cave and use a patch to tweak the difficulty or use a guide or something), but A Link to the Past and Link’s Awakening (DX) are more modern so maybe I’d have better luck with those (although that’d probably make it even harder to go back and play the original). Super Mario Bros. is still pretty accessible if you’re into platformers as long as you know about how to continue from the Start screen, but The Legend of Zelda is pretty brutal when it comes to not holding your hand, so it’ll take a mentality shift to fully enjoy it. Might want to go back and read some contemporary coverage of it, since that always gives me perspective.

Also Skull Kids don’t like adults, as one Kokiri will tell you.

Ah, I missed this, but when I returned to the Lost Woods as Adult Link I was happy to see that they were there, just hostile, and Navi explained about how they didn’t like adults. I don’t mind being heckled as long as they’re okay!

however try going at night

…of course Well, I’m still missing some Skulltulas in both the Lost Woods and Kokiri Forest, but I at least managed to rustle up a few more (the one on said elevated part of the room in the Lost Woods as well as one on top of a building in Kokiri Forest), and overall I managed to juuust scrape together 50 Gold Skulltula Tokens. There were two weird ones, though, which I suspect I may have collected in an unintended manner. In Dodongo’s Cavern, after running to and fro I finally spotted one in that corridor with the teeny Dodongos (not sure of their actual name). It was up on a ledge that I couldn’t climb up to nor reach directly with my hookshot, so I ended up dragging one of the statues used for pressing a button all the way over, backflipping onto it (since you can’t climb on them), and then pulling myself onto the ledge. The other one was more straightforward: in Zora’s River, there’s a Skulltula on the wall a bit before the entrance to Zora’s Domain, and while it wasn’t there as Young Link and was out of range of the Hookshot as Adult Link when using a normal approach, but by precisely standing on top of the fence I was juuuuust able to reach.

The theme and level design mesh well; it feels like a rich chateau that’s haunted by centuries of overgrowth.

It really does have quite the atmosphere! Amazing what they were able to pull off with the relatively primitive hardware—just goes to show that it’s art direction that matters, not polygon counts or texture resolution.