‘Perfect Japan’ posts spark Gen Z social media backlash

By AFP
March 23, 2026


Japan welcomed a record number of tourists last year

 - Copyright AFP Richard A. Brooks

Take an everyday video on any suburban transport network, add anime-style music and a rosy filter, and it’s suddenly a scene from the Japanese holiday of your dreams.

That’s the “Japan effect”: a Gen Z social media trend satirising the often-romanticised image of the Asian country, which welcomed a record number of visitors last year.

Residents of Kyoto and other tourist hotspots have expressed exasperation with selfie-taking crowds, and now an online backlash against Japan fever is growing.

The short video posts on platforms like TikTok show how even just the words “Tokyo, Japan” with a cherry blossom emoji can make an otherwise banal street scene more appealing for some users.

“The point is to make fun of Japan’s ‘cute’ image online, with all its cliches and stereotypes,” 25-year-old French YouTuber Rocky Louzembi, who analyses internet culture, told AFP.

Along with the chronically weak yen, the booming popularity of anime and game franchises such as Pokemon is drawing tourists to the nation.

But some people take their love of Japan too far, said Louzembi, who goes by the handle rockylevrai.

To describe the phenomenon, he used the slang word “glazing” — to excessively praise something.

A “Japan glazer” is “someone who puts everything that comes from Japan on a pedestal, while disparaging things that come from their own country”, Louzembi said.

– ‘Not that clean’ –

Japan logged a record 42.7 million tourist arrivals in 2025, despite a steep fall in Chinese visitors in December due to a diplomatic row.

Many visitors post online about their trip — making pilgrimages to real-life locations from cartoons or joking about spending $1,000 on flights just so they can eat a $1 convenience store rice ball.

“The ‘Japan’ portrayed in an anime world is often quite different from how Japanese society is”, said Marika Sato, a 29-year-old who works in marketing in Tokyo.

For instance, many women have experienced groping, said Sato, a contributor to “Blossom The Project”, an Instagram account focused on Japanese social issues.

Graphic designer and fellow Blossom contributor Maya Kubota, 28, said that she appreciates people liking Japan and wanting to visit.

But over-the-top comments such as “Japanese people are next level” give her an “icky vibe”, Kubota told AFP.

Some of the online Gen Z pushback focuses on the exaggerated idea that Japan’s streets are so spotless people don’t even have to wear shoes.

“Japan is clean but not THAT clean,” joked a US couple who post social media content about the country under the name The Hitobito — showing off their dirty white socks after a real-life experiment.

– Viral effect –

Japan’s tourist boom has forced some authorities to take action.

A cherry blossom festival boasting a highly Instagrammable view of Mount Fuji was cancelled this year after residents complained of overtourism.

“People associate Japan with carefully composed visuals,” said Seio Nakajima, a professor in the Graduate School of Asia-Pacific Studies at Waseda University.

That could be because of the detailed, beautiful backgrounds in anime, or because of a deeper “cultural tradition of emphasising form”.

“If people focus on form rather than meaning, it becomes easier to go viral. Because you don’t need to think,” Nakajima told AFP.

Japan’s formalities — from the complexity of polite language to extreme attention to detail in packaging or wrapping — may surprise visitors, he said.

But “Japan is not always clean and aesthetic. That’s only part of the reality.”

Despite the backlash, tourists in Tokyo’s busy Tsukiji market told AFP that the country had lived up to their expectations.

“In Russia, it’s very popular to hype Japan,” said Tatiana Mokeeva, 25.

When asked if posts about Japan could be unrealistic, she said: “To tell the truth, no… I love all about Japan.”
FacebookTwitterLinkedInEmailShare

No ‘silver bullet’ for video game age restrictions: PEGI chief


By AFP
March 23, 2026


Vastly popular existing titles like Fortnite will not be affected by a new PEGI classification - Copyright AFP/File Ina FASSBENDER

Kilian FICHOU

The head of Europe’s video game rating system, PEGI, has warned against supposed “silver bullet” child protection solutions such as age verification, in an interview with AFP.

A new set of PEGI (Pan-European Game Information) age ratings, coming into force from June, will take into account factors including in-game purchases, incentives to constantly revisit games or the ability to limit in-game messages from strangers.

It had taken “a couple of years” for PEGI to work out the new classification, its director general Dirk Bosmans told AFP.

The games sector has in recent years been the subject of debate, including over allegedly addictive mechanics such as “loot boxes” — virtual items purchasable for real money that contain a random in-game reward.

PEGI’s new ratings will not apply to games released before June this year — even the most widely played titles, such as “Fortnite” or “League of Legends”.

In future, “we will have to work out a plan of attack, an approach to live service games,” Bosmans said, “especially games that will continually provide new updates”.

Introduced in 2003, PEGI is the only media age classification system harmonised across European countries, its chief noted — although Germany has its own ratings.

As a self-regulatory mechanism by the games industry, its rules are applied by major console makers Nintendo, Sony and Microsoft, as well as by Google on its app store.

Apple has its own age rating system, while the dominant PC gaming platform Steam — based in the US — has not implemented one.



– ‘Regulatory pressure’ –



PEGI has updated its approach in part in response to growing “regulatory pressure” within the European Union, Bosmans said.

Even as the EU has tightened digital regulation in recent years, member states are taking their own steps — including a draft law in France barring under-15s from social media, which the government has warned would cover some online games with social aspects, such as “Roblox”.

If passed, the law will require all users to prove their age from 2027.

While automated online verification “sounds like it’s going to fix everything… data protection organisations are very concerned”, Bosmans said.

“We first need to have a really good conversation before we start deciding on where to apply it.”

He added that companies in the sector have welcomed the updated PEGI classifications.

“They understand that by making PEGI better and stronger, they are better protected against lack of nuance, quick fixes,” Bosmans said.



– Parents needed –



Bosmans also spoke out against full-on bans of games for children below a certain age — as mooted by French President Emmanuel Macron last month ahead of an expert inquiry.

“A ban is not very nuanced. It’s not very proportionate, no matter for what you apply it,” he said, recalling that PEGI was created to avoid just such a scenario in the early 2000s.

What’s more, in Australia — where social media has already been banned for under-16s — “there is now concern that kids are primarily busy with trying to circumvent the rules, sometimes with the help of their parents,” Bosmans said.

“You can try all kinds of technical or legal methods to enforce PEGI ratings. If in the end parents decide, no, my 13-year-old is going to play this 16 (rated) game, it doesn’t change anything,” he added.

“Thinking that you can do it without the parents is the biggest mistake you can make.”

Jury at US social media addiction trial reports ‘difficulty’ in finding consensus


By AFP
March 23, 2026


Mark Lanier, a lawyer for plaintiffs in the social media addiction trial in Los Angeles, speaking with journalists on Friday, March 20 - Copyright AFP Apu GOMES

Romain FONSEGRIVES

The jurors in a landmark social media trial signaled Monday that they could not reach a consensus against one of the two defendants, Meta and YouTube.

“The jury has difficulty coming to a consensus regarding one defendant, do you have any advice on how to move forward?” the jurors told Judge Carolyn Kuhl, according to a note she read out loud.

Kuhl responded by asking the jurors to continue their deliberations.

“If you are unable to reach a verdict, the case will have to be applied before another jury selected in the same manner and from the same community from which you were chosen, and add additional cost to everyone,” she told the jurors.

The development came after the jury’s first full week of deliberations ended Friday with the panel sending the judge a query related to calculating damages in the case, which is expected to set a precedent for thousands of similar suits in the United States.

That indicated enough jurors agreed that one or both of the tech platforms was negligently or harmfully designed and users should have been warned, according to verdict forms.

The lawsuit is one of hundreds accusing social media firms of luring young users into becoming addicted to their content and potentially suffering from depression, eating disorders, psychiatric hospitalization and even suicide.

– ‘Negligent’ designs –

Internet titans have long shielded themselves with Section 230 of the US Communications Decency Act, which frees them of responsibility for what social media users post.

But this case argues that the firms are responsible for defective products, with business models designed to hold people’s attention and to promote content that can harm their mental health.

The verdict could turn on the question of whether familial strife and other real-world trauma, or rather YouTube and Meta apps such as Instagram, are to blame for the mental woes of the woman who filed the suit.

A 20-year-old California woman identified as Kaley G.M. testified at the trial that YouTube and Instagram fueled her depression and suicidal thoughts as a child, telling jurors that she became obsessed with social media, starting with YouTube videos, when she was six.

Under cross examination, however, Kaley also talked about feeling neglected, berated and picked on by family members.

A jury form given to jurors asks the panel to decide whether Meta or YouTube should have known their services posed a danger to children or if they were negligent in design.

If so, jurors are to decide if Meta or YouTube were “substantial factors” in causing Kaley’s woes and how much they should pay in damages.

Meta awaits verdict in New Mexico child safety trial


By AFP
March 24, 2026


New Mexico Attorney General Raul Torrez filed suit in 2023 against Meta, alleging the company failed to protect children from sexual abuse, online solicitation and human trafficking - Copyright GETTY IMAGES NORTH AMERICA/AFP Jemal Countess

A New Mexico jury began its first full day of deliberations on Tuesday in a trial where social media giant Meta is accused of endangering children by making them vulnerable to predators.

The state of New Mexico is seeking billions of dollars in penalties in one of two major US cases against the company now in jury hands.

A separate jury in California is weighing whether Meta and YouTube should be held liable for harms caused to children on their platforms, including by making them addictive.

That case is considered a bellwether that could influence the outcome of thousands of similar lawsuits against social media companies across the United States.

The New Mexico jury began its work following closing arguments and a six-week trial involving testimony from 40 witnesses, including employees turned whistle-blowers, and hundreds of documents, reports and emails.

New Mexico Attorney General Raul Torrez filed suit in 2023 against Meta — parent company of Facebook, Instagram and WhatsApp — and CEO Mark Zuckerberg, alleging the company failed to protect children from sexual abuse, online solicitation and human trafficking.

Prosecution attorney Linda Singer told jurors in closing arguments that Meta’s algorithms had directed adults toward content posted by teenage users while the company concealed internal findings about the risks to young people.

“Meta failed to explain that the algorithm was designed to maximize teen time spent on the platform,” Singer said, according to the Albuquerque Journal. “Meta didn’t disclose the likelihood that the algorithm would introduce predators to teens, that it would recommend such sensational and harmful content.”

A Meta spokesperson said the state’s case was “sensationalist” and based on “cherry-picked” documents. “The State failed to prove its case,” the spokesperson said. “We’re focused on demonstrating our longstanding commitment to supporting young people.”

The state is seeking the maximum civil penalty of $5,000 for each of an estimated 221,000 New Mexico teenagers it says use Facebook and Instagram, a figure that is contested by Meta.

New Mexico’s attorneys must prove Meta violated the state’s Unfair Practices Act by misleading residents about the safety of its products for children.

The case, tried before First Judicial District Court Judge Bryan Biedscheid, is among the first involving social media platforms and child safety to reach a jury.

A second phase of proceedings in New Mexico is scheduled for May, when a judge will hear the state’s claim that Meta created a public nuisance and should fund programs to address alleged harms to children.

Data canary shows economy already suffering from Middle East war


By AFP
March 24, 2026


Higher energy prices are already affecting economies, survey data shows 

- Copyright AFP/File Tiziana FABI

Martine PAUWELS

Soaring fuel prices due to the Middle East war and snarled supply chains are already hurting businesses, business survey data published Tuesday showed.

Now in its fourth week, the war sparked by US and Israeli strikes on Iran has seen global oil prices soar by more than 40 percent as Tehran has effectively shut the Strait of Hormuz through which a fifth of oil and liquefied natural gas supplied flowed before the conflict began.

Economists have warned that if they persist higher energy prices could trigger a fresh surge in inflation and slow economic growth.

The Purchasing Managers’ Indices compiled monthly by S&P Global are the data equivalent of the canary in the coal mine.

They survey managers who have their thumbs on the pulse of businesses across many industries, and often reveal changes in business conditions months before official government data.

The latest batch of PMI surveys, which included the period since the war broke out on February 28, showed that businesses activity is already slowing and prices rising.

The initial reading for the composite US PMI dipped to an 11-month low of 51.4 points in March from 51.9 points in February, with services taking a hit while manufacturing edged higher.

A reading above 50 points indicates economic growth.

“The flash PMI survey data for March signal an unwelcome combination of slower growth and rising inflation following the outbreak of war in the Middle East,” said Chris Williamson, chief business economist at S&P Global Market Intelligence.

“Companies are reporting a hit to demand from the additional uncertainty and cost of living impact generated by the conflict,” he added.



– Stagflation threat –



Williamson said that the price component of the surveys indicated inflation rising back to around four percent “hinting at a growing risk of the US moving into an environment of stagflation.”

Stagflation is a period of little or no economic growth and high inflation, which poses a quandary for central bankers as raising interest rates to reduce inflation is a sure recipe to trigger a recession.

Meanwhile the eurozone PMI dropped to 10-month low of 50.5 in March, down from 51.9 in February, signalling a near stop in growth and weakening demand.

“The flash Eurozone PMI is ringing stagflation alarm bells,” said Williamson.

Analysts said the data is a warning signal.

“The risk is that the PMI data, which is a lead indicator, is the start of a wave of weaker economic data to come down the line,” said Kathleen Brooks, research director at XTB.



– Growth at near stop –



Christophe Boucher at ABN AMRO Investment Solutions said the impact of the war in Iran is visible in the PMI data in three ways: a slowdown in the growth in services, an increase in manufacturing prices and a degradation in the global outlook.

Services is by far the main largest economic sector, and in the United States the services PMI dipped to an 11-month low of 51.1.

In Europe it fell to 50.1, a 10-month low.

Germany however saw manufacturing output jump to a four-year high thanks in part a major public investment programme to boost the economy.

But “the problem with German industry is that it is extremely dependent upon access to fossil fuels” which raises concern that it will also face headwinds, said Christopher Dembik, investment strategist at Pictet Asset Management.

In France, the private sector registered its strongest contraction since October at 48.3 points.

The same slowdown trend was observed in PMIs in Britain and Australia.



– Inflationary spiral –



The survey found war-related shipping issues were a key cause of longer supply delivery times.

In the eurozone, input prices increased at the fastest pace since February 2023, with both manufacturing and services facing steeper inflation, due to higher energy prices.

In the United States, input prices rose at the fastest rate in 10 months, and companies passed the higher costs to clients, with selling prices jumping at the fastest rate in over three-and-a-half years.

ABN AMRO’s Boucher said it is important to watch for companies passing on higher costs to clients.

“What seems to be the most important to monitor, particularly in case of an extended conflict, is the risk of the transmission of inflation in the services sector which would signal the second-round effects,” he said.

While Iran and Israel traded strikes on Tuesday, there was still cautious optimism that talks US President Donald Trump evoked on Monday to justify postponing threatened attacks on Iranian energy infrastructure might lead to a deescalation.

“With the Ukraine conflict in 2022 it took five weeks to realize that it wasn’t temporary and was becoming structural,” Pictet AM’s Dembik told AFP, referring to the rise in energy prices triggered by Russia’s invasion of its neighbour.

“Today, with some twenty days of fighting, we’re still in the middle of the zone of uncertainty,” he added.

Jack Allen-Reynolds, deputy chief eurozone economist at Capital Economics, said “if energy prices remain high, worse could be to come.”

Op-Ed: Australia sets the bar high for energy and water for new data centres


By Paul Wallis
EDITOR AT LARGE

DIGITAL JOURNAL
AUSTRALIA
March 24, 2026


Cooling vents on data centres in Virginia. Researchers hope that storing data on glass will save energy - Copyright AFP/File ANDREW CABALLERO-REYNOLDS

The reasons for data centres using lots of power and energy are surprisingly simple and astonishingly primitive. So is the business logic.

Water is required for cooling systems. The data centres use evaporative cooling, which is hardly surprising. It’s just that they use so much water.

Energy is required for high levels of processing power and for cooling systems.

Quaint.

What’s required for all this power and energy is a little harder to pin down. It’s an open-ended, constantly exponentially expanding demand for money to pay for these things.

If you were trying to find the exact opposite of cost-effective planned system construction, this is it. There are a few things missing from this antiquated kitsch-based approach to essential services.

Energy efficiency is a 50-year-old best practice maxim in any electronic system design.

Water efficiency is a 5000-year-old basic best practice for human societies.

Cost efficiency is the basic best business practice.

Then there’s a rustic thing called “market efficiency”. This happens when your market doesn’t want to pay top dollar for “whatever”.

There are a few unexpected problems here. Does the wider market really need AI? Not really. It can do most of what it needs without it. Spending millions or billions and calling it “productivity” doesn’t wash with the hardheads. ROI matters. So does disrupting ongoing business for the sake of tech that will be obsolete in 10 years.

Let’s start with energy efficiency. If you know how a switch works, you probably also know that adding a few digits with dollar signs to turning a switch off and on doesn’t make a lot of sense. Running gigantic amounts of power to supply the answer to “Why am I a chicken?” isn’t necessarily a good use of resources.

If you happen to know that new tech like photonic electronics are already in the pipeline, do you really want to fund a techno-sandcastle? Sandcastles are traditionally built in the presence of large amounts of liquidity. They fall apart when they can’t handle the effects of liquidity. Poetic, huh?

Let’s not bash AI hype any more than required. It no longer needs mentioning. This tech at the consumer and business level doesn’t really do very much. Making searches a bit more efficient, great. Chatbots, so what? A turgid report nobody reads, who cares? AI screws up and produces a cluster of added costs and likely lawsuits, why do you bother? AI hallucinations, what the hell are you doing?

This is an investment?

Investment in what, may one make so bold to ask?

Let’s get real. AI is far more efficient functionally as a scientific asset. There it pays for itself and has metrics to prove it. The rest is just cosmetic dross. It’s a novelty act until further notice.

Water efficiency could be a real killer. US water shortages alone could reach lethal levels by 2071. The rest of the world also has major water issues, depending on regions.

Global energy costs are also getting worse. The world’s most mismanaged and mythologized sector is making a dog’s breakfast out of cost projections. Grids handle X amount of energy. If you want to expand, the added costs will be astronomical. See any issues?

This is where Australia’s new rules, which equate to “BYO water and power,” come in. Australia is now the second-largest developer of data centres. Development proposals will need to meet these standards to proceed.

More annoyingly, all of these energy and water problems are fixable with a good initial design. The word is “sustainable”. It’s a business principle, not some recycled hippie buzzword. If you want to be in business next week, you MUST address sustainability.

Why so much current for processing? What’s wrong with microvolts? Nano systems use far less power. Why stick a gigantic, expensive straw in the world’s water supply when you could simply recycle the water to the point of a single metric?

Water and energy are not negotiable. Nor is proper costing. Take the hint.

__________________________________________________________

Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.

Countries most at risk from AI-led cyberattacks revealed



By Dr. Tim Sandle
SCIENCE EDITOR

DIGITAL JOURNAL

March 22, 2026


The number of hacks has been increasing worldwide. — © AFP/File Noel Celis

According to a March 2026 report on the effects artificial intelligence has on cybersecurity, Bosnia and Herzegovina is at the biggest risk because of fast-paced AI adoption.

The report comes from a cybersecurity company called Check Point. The report compares countries’ digital defences to the levels of AI-fuelled cyber threats.

The study looked at 38 countries, evaluating them by three key criteria, influencing AI’s impact on cybersecurity. The report considered the extent of AI integration (in %) and the cybersecurity levels, which were quantified by policies, infrastructure, cyber incident response and crisis management. These factors were compared against the cybersecurity threat level that was calculated as % of exposure to the Botnet (device infection), Infostealer (personal information theft), banking trojans, ransomware, and mobile attacks. Lower Security Index highlights the countries less prepared for AI threats.

The countries with the biggest gap between AI adoption and security

Country	AI Diffusion	Cybersecurity Policy	Cybersecurity of Critical Information Infrastructure	Cyber Incident Response	Cyber Crisis Management	Total Cyber Threat Exposure (%)	Security Index
Bosnia And Herzegovina	20%	0	0	0	0	14%	22
Kuwait	19%	40	0	43	0	19%	30
Qatar	38%	60	0	21	56	6%	39
Jamaica	22%	40	0	57	33	15%	40
Costa Rica	27%	40	25	64	22	6%	47
Argentina	20%	60	50	57	22	13%	51
South Africa	21%	80	50	64	0	13%	54
Panama	22%	80	25	43	56	12%	55
New Zealand	41%	60	25	79	44	7%	64
Sweden	33%	20	75	100	22	10%	72

From the above table, Bosnia and Herzegovina stands out as the country where AI adoption outpaces cybersecurity measures the most, with 1 in 5 organizations using artificial intelligence in public, business and technological sectors. The country ranks low for digital policies and crisis management, scoring 0 for all security measures that protect from cyber threats. Around 14% of the country’s infrastructure is vulnerable to hacking, and Botnet attacks are the most common.

In second place is Kuwait, with the largest exposure to cyber threats in the top 10. Almost 20% of the country’s digital frameworks are susceptible to cyber attacks, while Kuwait is broadening the AI use to 1 in 5 of business and tech companies. Kuwait updated its digital policies sooner than Bosnia and Herzegovina, but it doesn’t have proper cybersecurity for its critical information infrastructure or cyber crisis management.

Qatar integrates AI in its tech and business sectors more than any other country in the top 5. Around 38% of the organizations use it in their work, while the country doesn’t have records of cybersecurity measures for critical infrastructure. Around 6% of Qatar’s infrastructure is open to cyber threats, and ransomware is used the most to attack the country’s residents.

Jamaica also implements AI technologies fast, while not preparing sufficient cyber defences. Around 22% of the tech and business sectors already adopted AI, but the country hasn’t passed matching regulatory policies, scoring only 40/100. Jamaica also shows low cyber incident response (57/100), while 15% of all digital infrastructure in the country is at risk of hackers.

Costa Rica rounds up the top 5, with over a quarter (27%) of tech, business and private organizations using AI for their tasks. The country is not as exposed to cyber threats right now as other countries on the list, with only 6% of digital systems being vulnerable. The fast pace of AI integration makes Costa Rica a desirable target for hackers in the near future, as it shows almost no cybersecurity measures prepared for critical infrastructure.

A pattern from the above analysis is where AI is now both a gateway for a cyberattack and a tool. Botnets were dangerous before, and AI has made them fully autonomous. AI agents handle reconnaissance, launch infections, and adapt to defences, compressing multi-day operations into minutes.

Musk has the world’s most powerful supercomputer, but it is also the most energy hungry


By Dr. Tim Sandle
SCIENCE EDITOR

DIGITAL JOURNAL

March 22, 2026


World's richest man Elon Musk is on track to become the world's first trillionaire - Copyright GETTY IMAGES NORTH AMERICA/AFP/File Christopher Furlong

A report on AI supercomputers indicates that Elon Musk’s xAI Colossus is the most powerful AI machine on the planet. No company has invested more heavily in AI computing than Meta, which owns three of the top 10 AI supercomputers.

With artificial intelligence now consuming as much electricity as mid-sized cities, the report, data infrastructure provider TRG Datacenters, reveals the world’s most powerful AI infrastructures and the companies behind them.

To compare supercomputers built with different hardware, the research converted every system’s chips into H100 equivalents, a common unit based on NVIDIA’s most widely used AI processor. Alongside raw computing power, the report also calculated how much electricity each cluster consumes per 10K H100 equivalents, showing which systems deliver more computing output per megawatt of power. The study is based on Epoch AI’s global GPU cluster database.

The 10 most powerful AI supercomputers in the world

Name	H100 equivalents	Owner	Country	MW per 10k H100-eq	Power (MW)
xAI Colossus Memphis Phase 3	275,796	xAI	United States of America	12.78	352.40
Meta 100k	100,000	Meta AI	United States of America	14.27	142.70
OpenAI/Microsoft Goodyear Arizona	100,000	Microsoft,OpenAI	United States of America	14.27	142.70
Oracle OCI Supercluster H200s	65,536	Oracle	United States of America	14.27	93.50
Tesla Cortex Phase 1	50,000	Tesla	United States of America	14.27	71.30
Lawrence Livermore NL El Capitan Phase 2	44,143	US Department of Energy	United States of America	7.93	35.00
CoreWeave H200s	42,000	CoreWeave	United States of America	14.27	59.90
Meta GenAI 2024b	24,576	Meta AI	United States of America	14.27	35.10
Meta GenAI 2024a	24,576	Meta AI	United States of America	14.27	35.10
“Jupiter, Jülich”	23,536	EuroHPC JU,Julich Supercomputing Center	Germany	7.65	18.00

xAI Colossus in Memphis is the world’s most powerful AI supercomputer by a wide margin. Its 275K+ H100-equivalent chips are nearly three times what Meta and OpenAI/Microsoft could put together in their own clusters. The whole system runs on 352.4 MW of power, about as much electricity as a city of 250K people uses on a typical day. And unlike most commercial clusters in the ranking, Colossus uses less power for every unit of compute, at 12.78 MW per 10K H100 equivalents.

Meta’s 100K cluster comes in second with exactly 100K H100 equivalents, one of two systems to reach that figure. The cluster needs 142.7 MW to operate, about 60% less than Colossus, although this is because it’s a third of its size in computing terms. Meta built this system specifically to train its Llama family of AI models, and its scale shows how aggressively the company has been pushing into the AI race.

The OpenAI and Microsoft joint cluster in Goodyear, Arizona, ties with Meta for second place in raw computing power at 100K H100 equivalents. It also requires the same 142.7 MW of power, which makes it identical to Meta’s system on paper. In practice, though, this facility is the physical infrastructure behind the supercomputing partnership that brought OpenAI’s ChatGPT and Microsoft’s Copilot to billions of users.

Oracle’s H200 supercluster ranks fourth with 65K+ H100 equivalents. This system is actually built on NVIDIA’s newer H200 chips, which offer faster memory speeds than the standard H100. The cluster consumes 93.5 MW of power and runs at the same 14.27 MW efficiency ratio as the two Meta and Microsoft systems above it. Oracle has been quietly building out one of the largest AI cloud infrastructures in the world, and this cluster represents its most powerful system.

Tesla’s Cortex cluster rounds out the top five with 50K H100 equivalents and a 71.3 MW power usage. Unlike every other system in the top five, Cortex wasn’t built to sell AI computing to other companies. Tesla runs it entirely in-house to train its Full Self-Driving software, processing billions of miles of real-world driving footage to teach its cars how to move on roads.

The top 20 clusters in the ranking together need over 1,200 megawatts just to stay on. By adding in cooling, this adds 30 to 50 percent on top of the base power usage.

MAX IS T.I.N.A.

Russia’s Max: The unencrypted super-app being forced on citizens


By AFP
March 23, 2026


The rollout has raised concerns that Moscow will use Max to surveil its citizens - Copyright AFP SEBASTIEN BOZON

Russia is pushing its Max messenger — a social media platform without encryption — onto its citizens with a massive promotion campaign and the simultaneous blocking of Whatsapp and Telegram, the country’s two most popular messenger apps.

The rollout has raised concerns among critics and digital rights groups that Moscow will use Max to surveil its citizens and further cut digital links to the West.

“Any data that passes through this application can be considered to be in the hands of its owner, and in this case, the hands of the Russian state,” cybersecurity researcher Baptiste Robert, CEO of the French company Predicta Lab, told AFP.

Launched in 2025 by Russian social media giant VK, the app has been compared to China’s WeChat, combining social media and messaging functions with access to government services, a digital ID card system, banking and payments.

It is not officially mandatory, but the authorities are making it clear that life without Max will become increasingly hard.

President Vladimir Putin has touted it as a more “secure” platform that meets Russia’s demand for “technological sovereignty.”

Moscow has been pushing that agenda for years.

“This is the culmination of policies aimed at creating a sovereign internet,” Marielle Wijermars, an associate professor of internet governance at Maastricht University told AFP.

“Russia wants to restructure the internet to better control what is published” including “by migrating all Russians to platforms that are more state-controlled,” she added.

– ‘Forced’ to download –

Max has been pre-installed on phones and tablets sold in Russia since September.

The design is familiar and resembles Telegram, offering private messages, public channels and cute stickers.

Unlike Telegram and Whatsapp, it is also on Russia’s “white list” of approved digital services that stay online during the increasingly common forced internet blackouts that Moscow says are necessary to thwart Ukrainian retaliatory drone attacks.

Initially only available to users with a Russian or Belarusian SIM card, the app is now available in English and to those with phone numbers from 40 other countries — only those Russia deems “friendly,” like Cuba, Pakistan and ex-Soviet republics in Central Asia.

It is not available in the European Union — or Ukraine.

That has not stopped Ukrainian President Volodymyr Zelensky vowing to infiltrate the messenger.

One of the reasons Russia wants to ditch Telegram is because it has become a platform used by Ukraine to recruit Russians for sabotage attacks, including assassinations.

Inside Russia, opinions are split.

“You can send messages, photos and videos. What more do you need?” said Yekaterina, a 35-year-old dance teacher.

Irina, a 45-year-old doctor, however, complained she has been “forced” to use Max for school activities for her children and to access the government’s official online portal, Gosuslugi, where her patients make appointments.

She plans to “buy another SIM card to download Max on another phone.”

Large businesses have been accused of forcing employees to download the app and schools have migrated all communication with parents to the platform.

At the same time, celebrities and popular bloggers are moving their content to Max.

Dmitry Zakharchenko, founder of the Russian analytics agency GRFN, has compared the “aggressive” campaign with Soviet propaganda billboards.

The carrot-and-stick approach has driven downloads — more than 100 million users in March, according to the service.

– ‘Being watched’ –

The launch of Max comes years into Russia’s political and technological campaign to develop a “sovereign internet”, less reliant on — and vulnerable to — foreign services.

Russian telecoms regulator Roskomnadzor and the security services have enjoyed growing powers to monitor and block sites they deem dangerous.

Unlike Telegram and Whatsapp, Max does not use end-to-end encryption and its terms of use state that user data is stored exclusively on services in Russia.

Varvara, a 35-year-old interpreter said she was not worried about that as she was not a “foreign agent” and had nothing to hide — referring to a label used by the Kremlin to target critics.

Even so, she has ditched Max in favour of IMO, a less popular US-made app that has encryption.

Scientist Alexandra, 32, refuses to download Max “out of contrariness” to its heavy-handed promotion.

“We’re already being watched everywhere,” she added, dismissing the privacy concerns.

But another resistant user — Natasha, 48 — shows the general feeling of resignation when it comes to the future of the app in Russia.

“Sooner or later, there will be no alternative.”

Why most robot demos don’t hold up in the real world


By Digital Journal Staff
March 23, 2026


Photo by Simon Kadula on Unsplash

Robots look great in a lab. The floors are clean, the lighting is controlled, and tasks unfold exactly as planned.

Even if that task is taking a Queen dance break.

It’s easy to watch robotics demos and assume the future of automation is just a matter of scaling what already works.

In a conversation at CES 2026, Mikell Taylor, head of robotics strategy at General Motors, makes the case to Ani Kelkar, partner at McKinsey’s Boston office, that this version of the future is incomplete. Robot talk has historically and colloquially been about replacing people with fully autonomous systems.

Rosie on The Jetsons, anyone?

Instead, it’s time to think about how humans and machines work together in environments that are unpredictable, high-pressure, and far less forgiving than any demo. For instance, in manufacturing, on a factory floor.

This means designing systems that can handle real-world variability and building robots that are tested where the work happens. It also means acknowledging that the hardest problems are not isolated to one layer of the technology. Progress depends on getting the whole system to work together, from hardware and software to how people use it day to day.

In the interview, Taylor focuses on what it takes to deploy robotics in real environments, from how teams work with automation to how systems are designed and tested.

Here are three key moments from the conversation.

On the myth of fully autonomous “lights-out” manufacturing and the necessity of human collaboration

“The nuance that can’t be overlooked is that success will rely on how well people and automation work together. I am not a believer in lights-out manufacturing. I’ve worked long enough in robotics to know that stage is a wonderful North Star, but it never becomes a reality…The roles that the people perform may be different, but they’re still going to make the orchestra play together to get the kind of manufacturing capability and capacity that we want.”

On the need to prioritize real-world testing over flashy laboratory demonstrations

“Fundamentally, it requires engineers to be willing to hear that their baby is ugly, and for them to take that baby out into the world as soon as possible and hear that. The longer something stays in a lab or in shiny videos, the less successful it’s going to be…. We need to treat it like any other equipment and say, ‘Let’s try this. What do you think? How can I make it better?’ We need to involve workers in the process and ensure they’re part of the design effort to create a product people want and love, not just a cool robot.”
On the greatest challenge facing the robotics industry

“You’ll hear some companies and some researchers say, ‘The software is a commodity. The problem’s in hardware.’ Some other folks will say, ‘The hardware is a commodity. Now the problems are in software.’ Neither is true. The problems are everywhere. The success lies in the fully integrated system that is co-designed top to bottom in the stack, physical and virtual.”

Read the full interview with Taylor here.