A SELL-OFF in Taiwan Semiconductor Manufacturing Company (TSMC) shares looks almost inevitable when trading resumes after a two-day typhoon break, during which global investors dramatically soured on the promises of artificial intelligence (AI).

Investors sold off US Big Tech stocks over the past couple of sessions, as lacklustre earnings from Tesla and Alphabet set off a reckoning on the AI hype. TSMC may play catch-up to the losses seen in its American Depositary Receipts, which are down nearly 6 per cent over the past two sessions.

The latest tech rout adds further pressure on the world’s largest contract chipmaker, whose record-breaking rally has been faltering since mid-July. A favoured AI play due to its cutting-edge chips and a stellar run of earnings, concern over pricey valuation and the risk of tighter US curbs on chip sales to China have poured cold water on the bullish momentum.

TSMC’s Taiwan listing has fallen nearly 10 per cent from its peak. Further declines would spell trouble for the local financial market given the stock’s outsized influence. It accounts for more than a third of the benchmark Taiex Index, and outflows will hurt the Taiwan dollar, which is hovering near the lowest since 2016 against the greenback.

TSMC’s recent drop could be due to profit taking while “whispers of a potential slowdown in the AI investment boom might also be at play”, said Manish Bhargava, chief executive officer at Straits Investment Management. “The wider context of AI momentum can’t be dismissed. The burning question is – is the AI rally running out of steam?”

A TSMC spokesperson said there has been no impact from the typhoon as at Thursday (Jul 25).

Taiwan’s US$2.5 trillion stock market was shut on Wednesday and Thursday as the deadly Typhoon Gaemi approached after inundating the Philippines. The last time that Taipei was shut for two days in a row for a typhoon was in 2016. BLOOMBERG

THE developer of an innovative job matchmaking app helping Japan tackle its severe labour shortage goes public on Friday (Jul 26) following the nation’s largest initial public offering (IPO) so far this year.

Timee is listing on the Tokyo Stock Exchange’s Growth market section at an expected market value of about 138 billion yen (S$1.2 billion). That makes the Tokyo-based firm the biggest among 95 listings of application software companies in Japan since 1994, according to data compiled by Bloomberg.

With a chronically low birth rate, an ageing population and a strict immigration policy, the job crunch in the world’s fourth-largest economy is acute. The Bank of Japan’s latest Tankan survey of business sentiment showed shortages are on the rise, particularly among non-manufacturers, which are experiencing their worst manpower constraints in more than 30 years.

The service offered by Timee is aimed at filling gaps with so-called “spot work” that matches employers with people seeking jobs for specified periods of time and locations. Such arrangements offer greater flexibility than traditional part-time jobs as well as gig work such as ridesharing and food delivery.

The company will not get any of the roughly US$300 million in proceeds from the IPO, as the shares are being sold by holders including Ryo Ogawa, the company’s chief executive officer, and Internet advertising firm CyberAgent. Still, going public may help the company boost its profile among clients, jobseekers and investors.

“We want to offer our service to broader industries such as hotel, nursing care, childcare, and manufacturing,” Ogawa said on Tuesday. “We will aggressively accelerate investment as we gain credibility and a sense of security as a public company,” he said, adding that future mergers and acquisitions are possible.

Most of the startup’s client companies are in the logistics, catering and retail industries, Ogawa said. While spot work is still a new concept, Timee has already drawn competitors, with Mercari launching a similar service in March and Indeed owner Recruit Holdings planning its own offering for later this year.

Clarence Chu, an analyst at Aequitas Research, said Timee “appears to be a leader in a yet untapped and underpenetrated space”. The company’s stock “could likely end up trading at a premium to its entire peer set”, he wrote in a note on Smartkarma, adding that “over its relatively short operational history, Timee’s growth has been exponential”.

The shares were sold at 1,450 yen, the top of the expected range. In another positive sign, Japanese technology listings that raised at least US$200 million over the past ten years rose 15 per cent on average in their first day of trading, data compiled by Bloomberg show. BLOOMBERG

GOOGLE DeepMind, Alphabet’s artificial intelligence (AI) research division, said it has made strides in solving complex math problems, an area that remains challenging for today’s AI programmes.

On Thursday (Jul 25), Google rolled out AlphaProof, which specialises in math reasoning, and AlphaGeometry 2, an updated version of a model focused on geometry that the company debuted earlier this year. The programmes aced four of the six problems featured in the International Mathematical Olympiad, an annual competition in which students tackle topics such as algebra and geometry, Google said.

In the AI industry, where comparison between offerings is difficult, solving math problems has become a key proof point. That’s because large language models, which are trained on vast amounts of written text, tend to be biased towards linguistic rather than mathematical intelligence. While computers are good at numbers and traditional calculations, word-based math problems fall outside of these norms and require more sophisticated reasoning skills.

While AI tools are becoming more proficient at chatting naturally or producing images, they often struggle with problems that require planning or take multiple steps to solve. But Google and its competitors have not given up. The company’s biggest rival, OpenAI, has also been working on new reasoning technology, Bloomberg has reported.

AlphaProof evolved from Google AI programmes that have excelled at complex strategy games such as chess, shogi and Go, Google said. A DeepMind programme famously beat one of the world’s top Go players in 2016.

Large language models have a tendency to hallucinate, or deliver incorrect information in convincing fashion. Google said it sidestepped that challenge by using its AI to translate math problems into technical statements, or what it called “formal language”.

BT in your inbox

Start and end each day with the latest news stories and analyses delivered straight to your inbox.

Sign Up

Sign Up

Another issue for AI systems in mathematics is the lack of available training data, unlike chatbots, which can glean information from vast troves of text online. As Google’s AlphaProof model successfully solves problems, its code is updated, allowing it to tackle ever more difficult challenges, the company said.

The company also released an improved version of its AlphaGeometry AI model, which it said was able to solve 83 per cent of all historical geometry problems included in the International Mathematical Olympiad, spanning the last 25 years.

But Google’s researchers also said that AI is far from being able to replace human mathematicians with its problem-solving capabilities. “Even in the fullest ambition of what we are trying to do, I think we are aiming to provide a system that can prove anything,” said David Silver, Google DeepMind’s vice-president of reinforcement learning. “But that’s not the end of what mathematicians do.”

Silver said DeepMind’s AI models are more akin to slide rules or calculators: powerful computational tools that might one day help humans come up with mathematical proofs. But what the AI systems lack is imagination. “Mathematicians pose interesting problems,” he said. BLOOMBERG

ESSILORLUXOTTICA confirmed Meta Platforms is interested in buying a stake in the world’s biggest eyewear maker.

Chief executive officer Francesco Milleri said that the US tech giant intends to become a shareholder in the French-Italian group potentially building on a years-long partnership to develop smart glasses.

“We are proud that a company that knows us very well, after years of partnership, is convinced that our company can grow and make much better in the future,” he said on Thursday (Jul 25).

Speculations emerged earlier this month that the owner of Facebook is considering taking a stake of up to 5 per cent in EssilorLuxottica. Milleri said the maker of Ray-Ban and Oakley sunglasses is not planning a capital increase dedicated to Meta and the US firm would buy shares on the market if it wants to become a shareholder.

He did not elaborate on the size of the stake Meta may buy and on the timing of the purchase.

The world’s biggest manufacturer and retailer of eyewear posted an adjusted net income of 1.8 billion euros (S$2.6 billion) in the first half, 10.6 per cent higher compared to the same period last year at constant exchange rates and higher than consensus. Revenues were in line with analysts’ forecasts at 13.3 billion euros.

Performance in Europe, Middle East and Africa, where revenue jumped 7.9 per cent in the second quarter compared to the same period a year prior, offset lower growth in North America, held back by a decline in comparable-store sales at Sunglass Hut, the company said on Thursday.

EssilorLuxottica and Meta have been co-operating on smart glasses for a few years. In 2021, Meta unveiled its first Ray-Ban smart glasses set on the Wayfarer frame to let users take photos and videos, listen to music and answer calls. Newer glasses incorporate MetaAI, an artificial intelligence (AI) assistant based on its Llama AI model.

“We spend a lot of time together,” Milleri said. We have “a strong relationship with Mark Zuckerberg and the top management of Meta”. BLOOMBERG