From toilets to MSG: Four unlikely winners of the AI boom
Believe it or not, Toto, the Japanese toilet and bidet maker, has become one of the more improbable stock success stories of the AI boom.
The company produces electrostatic ‘chucks’, components critical to memory chip production, and when activist fund Palliser Capital took a stake last month and called it “the most undervalued and overlooked AI memory beneficiary”, the shares jumped over five per cent.
They’re up 54 per cent over the past year. Not bad for a bidet.
Toto’s surge begs the question: who else? The AI boom, it turns out, has a peculiar habit of enriching companies with no direct connection to technology whatsoever.
While markets obsess over Nvidia’s guidance or the latest AI model, a seasoning firm, a paint manufacturer and a printing press have positioned themselves at chokepoints in the semiconductor supply chain. Here are the other three worth knowing about:
The MSG maker
Ajinomoto is best known for producing MSG, or monosodium glutamate, the flavour enhancer that has seasoned billions of bowls of ramen, and endured decades of nutritional suspicion.
The firm’s name translates as ‘essence of taste’. It is not, at first glance, a tech company.
In the 1990s, however, Ajinomoto’s researchers found that a by-product of their amino acid process made a remarkably effective electrical insulator.
They developed it into something called Ajinomoto build-up film (ABF), a material used in the packaging of advanced CPUs and GPUs, sitting between chips and making sure the copper wiring used in processors doesn’t short out and take the whole thing with it. Unglamorous work, but absolutely essential work.
Ajinomoto now holds over 95 per cent of the global market for this material. Every AI chip running in every data centre requires it.
The company has already spent 25 billion yen expanding its facilities outside Tokyo over the past two years, and president Shigeo Nakamuro, who fittingly spent his career in the electronics division rather than in the flavouring business, has pledged to invest at least the same again by 2030.
He has targeted a 50 per cent increase in production capacity for the firm, and is forecasting annual growth in electronics of over 10 per cent through the decade.
The stock is up 31 per cent year to date. For a firm whose name appears on seasoning packets across Asian supermarkets, it’s quite the subplot.
There is one cloud, however. A Berkeley startup, Thintronics, funded partly through the US chips act, is trying to develop a competing material and dethrone Ajinomoto in what amounts to a geopolitical supply chain project.
The paint company
This one asks for a small leap of faith. Sherwin-Williams has been selling paint for 150 years.
It has nearly 5,000 shops across the US, and its fourth results, posted earlier this year, showed record full-year sales and record earnings per share.
Chief executive Heidi Petz guided 2026 earnings to between $11.50 and $11.90. The stock is up 14 per cent year to date. None of that, by itself, screams AI.
But buried within the firm’s industrial division is a business producing thermal coatings, the kind used for the flooring of large-scale data centres.
AI data centres run significantly hotter than conventional ones; the hardware is denser, power higher, and the buildings that house them must meet increasingly demanding thresholds for heat management and fire safety.
Sherwin-Williams supplies the coating that help them do that, and the data centre construction boom has been a huge driver of its protective and marine segment.
It isn’t the firm’s whole story, but growth in that segment tied to infrastructure projects has contributed to what Mizuho, which raised its price target last month, called a business with continued “pricing discipline and cost execution”.
Sometimes, the picks and shovels play is quite literally paint.
The newspaper firm
Dai Nippon Printing (DNP) was founded in 1876. It prints things: books, magazines, packaging, the manga volumes that have driven Japanese pop culture for generations. It is, by any reasonable assessment, a legacy media business.
And yet, printing a paper and manufacturing a semiconductor chip have more similarities than one may expect.
Both are exercises in lithography, the transfer of a pattern onto a surface. DNP has spent 150 nailing that skill. And, as it turns out, that expertise maps onto making photomasks: the precision plates used to project circuit designs onto silicon during chip production.
As chips get smaller, the tolerances on those masks become inexplicably tight. The masks required for the current frontier of semiconductors require a level of precision that only a handful of companies can meet. DNP, the newspaper printer, is one of them.
In 2024 the firm was chosen as a subcontractor for Rapidus, Japan’s state-backed chipmaker targeting mass production by 2027.
DNP’s role is to develop and supply the photomasks Rapidus needs to print its chips.
Earlier this year, it went further, playing a part in Rapidus’ latest funding round as a strategic investor.
Shortly after, it released a new template, positioning itself for the new generation of even smaller chips. Mass production has been targeted for 2027. The stock is up 50 per cent year on year.
Supply chain winners
None of these firms set out to be AI stocks; Ajinomoto stumbled into semiconductors via acid chemistry; Sherwin-Williams built coatings that turned out to matter enormously to the construction of data centres; DNP applied a centry of precision printing to a problem that, structurally, wasn’t too different from the one it had been used to solving.
The AI boom has, to an extraordinary degree, been a story of mass investment within a very small handful of companies.
The Magnificent Seven, Alphabet, Nvidia, Amazon, Apple, Microsoft and Tesla, now account for roughly a third of the S&P 500 by market cap, a concentration that exceeds even the peak of the dot-com bubble in 2000.
JP Morgan calculated that AI-related stocks have driven around 75 per cent of S&P 500 returns since Chatgpt launched in late 2022.
Whether that boom will end badly has been consuming every investor for the past two years, and even more so in the past few months. What has also been debated, has been the infrastructure buildout it comes with.
Goldman Sachs estimates AI capital expenditure will rise a further 19 per cent in 2026. That money has to go somewhere, and an increasing amount it is flowing to the companies that supply the materials, coatings and various components without which none of it is physically possible.
Whatever becomes of Nvidia’s soaring stock price of the huge spending sprees of tech giants, the firms producing these vital components are not trading on AI hype based on estimated demand.
Instead, their success has hinged on actual sales and real purchases. And recently, those sales figures have looked very strong indeed.