Data points

Published
0
In an aerial view, the Meta El Paso Data Centre is seen in El Paso, Texas. Meta Platforms continues construction on its more than $10 billion Meta El Paso Data Centre in Northeast El Paso, Texas. Occupying approximately 1,000 acres, Meta has expanded the project to 1 GW of computing capacity, enabling the facility to support a growing range of AI workloads and the training and deployment of increasingly sophisticated AI models and technologies.—AFP
In an aerial view, the Meta El Paso Data Centre is seen in El Paso, Texas. Meta Platforms continues construction on its more than $10 billion Meta El Paso Data Centre in Northeast El Paso, Texas. Occupying approximately 1,000 acres, Meta has expanded the project to 1 GW of computing capacity, enabling the facility to support a growing range of AI workloads and the training and deployment of increasingly sophisticated AI models and technologies.—AFP

Catching lying students

At the California Institute of Technology, a little-known band of admissions officials is scrutinising the fonts on science-fair certificates, checking addresses on high-school transcripts and analysing signatures on awards. Known as the “fraud squad,” these investigators are in charge of vetting applications to the highly selective school, which admitted less than 4pc of its 11,400 applicants for the incoming class. Caltech created the team two years ago partly in response to a rising number of anonymous accusations of dishonest submissions. The four-person team inspects every admitted student’s application, taking nothing for granted. They scour the internet to confirm obscurities, contact competition officials and high-school counsellors, and use a video-interview program to suss out students’ research claims.

(Adapted from “The ‘Fraud Squad’ At Caltech Will Catch You If You Lie On Your Application,” by Roshan Fernandez, published on August 4, 2026, by the Wall Street Journal)

How to stop procrastinating

Putting things off exacts a hefty toll. One study published in 2013 ranked 22,053 people on a procrastination scale of one to five. Each increase of one point was associated with a drop of nearly $15,000 in annual earnings. Those who habitually engage in “self-regulatory failure”, as procrastination is also known, are more prone to delaying medical treatment. A study of 3,525 Swedish university students published in 2023 found that those who scored highly for procrastination were more likely to suffer debilitating pain. It gets worse. The same Swedish study also found procrastinators more prone to anxiety, stress and loneliness. A 2025 meta-analysis, published in Frontiersin Psychiatry and covering 88 studies involving 63,323 people in 17 countries, found procrastination and its resulting erosion of self-esteem to be a predictor of future depression.

(Adapted from “How To Stop Procrastinating,” published on July 31, 2026, by The Economist)

Bidding till broke

Seven-year-old Whatnot is one of the fastest-growing shopping platforms in America. It allows people to bid around the clock on trading cards, squishy toys, and instant ramen, all auctioned live in high-energy shows within the app. It has tapped into the addictive qualities of social media, with opportunities fed to you in an algorithm-controlled scroll. Often, users don’t search for a particular item, but instead hope to hit on something by being on the app at the right time. The design supercharges traditional shopping sites such as eBay and Amazon with high-pressure countdown auctions, often run by hosts who urge buyers to bid higher and faster. The most feverish shows are built around “breaks,” where a person buys an opportunity to win an unknown, potentially valuable, collectable. Whatnot said it added 20 million new accounts in 2025.

(Adapted from “The Live Shopping App Where Some People Bid Until They’re Broke,” by Hanna Krueger and Sarah Nassauer, published on August 4, 2026, by the Wall Street Journal)

AI and chip trends

Nvidia and its biggest customers — cloud giants such as Amazon, Google, Meta and Microsoft — still need each other. Nvidia designs and supplies chips; the hyperscalers build data centres using them. Yet both are preparing for a future in which they lean on each other less. A sign of impending separation came on August 10th, when Nvidia announced a partnership with six of Wall Street’s biggest investors, including BlackRock and Goldman Sachs, to “mobilise over $500bn” for AI infrastructure. The aim is to help customers, such as smaller AI labs and businesses that face steeper borrowing costs than Google or Microsoft, to find the vast sums needed to build data centres. Under the plan, the consortium will raise pools of capital from institutional investors and lend it to Nvidia’s customers at attractive rates to build infrastructure using Nvidia’s gear. Infrastructure is expensive: a large data centre costs around $50bn.

(Adapted from “Nvidia’s Great Silicon Showdown,” published on August 11, 2026, by The Economist)

Published in Dawn, The Business and Finance Weekly, August 17th, 2026

Opinion

Editorial

Real FBR reform
21 Oct, 2026

Real FBR reform

REFORM of the FBR has always been easier to promise than to deliver. Since the late 1990s, governments have tried...
Missing lawmakers
Updated 21 Oct, 2026

Missing lawmakers

Some lawmakers have reportedly been in detention since before Oct 11.
Tackling rabies
21 Oct, 2026

Tackling rabies

RABIES is a deadly but preventable zoonotic disease. On Monday, two more victims succumbed to it in Sindh, taking ...
Burnt bridges
Updated 20 Oct, 2026

Burnt bridges

If PPP wants the decision reversed, it should convince its ally in Islamabad, not ask the opposition to be ‘positive’.
Toll on trade
Updated 20 Oct, 2026

Toll on trade

Islamabad must find the right balance between its security priorities and the economic strain on KP’s residents.
French protests
20 Oct, 2026

French protests

SEVERAL European states have witnessed mass protests over the past few days, mostly focusing on socioeconomic...