Home Blog Page 41

People spend 15.6b hours on GenAI apps in first half of 2025

  • Downloads surge by 67% while consumer spending on IAPs doubles to $1.9b.
  • Asia leads the way now, particularly in fast-growing markets such as India and Mainland China, where downloads surge 80%, followed by Europe by 51% and North America by 39%.
  • DeepSeek, launched in January 2025, outperforms all other Generative AI apps—including ChatGPT—for new global downloads due to its strong uptake in Asia, the Middle East, and Africa.
  • ChatGPT remains the all-time download leader at 940m, followed by Google Gemini’s 200m and DeepSeek’s 127m.

The first half of 2025 has been nothing short of explosive for Generative AI apps around the globe, with users flocking to platforms that offer AI Assistance or AI Content Generation.

According to Sensor Tower’s State of AI Apps Report 2025, both the App Store and Google Play have seen Generative AI apps approach 1.7 billion downloads, while in-app purchases soared to nearly $1.9 billion.

What’s more impressive is that both downloads and IAP revenue are not just increasing—they’re accelerating, with half-over-half (HoH) growth rates hitting their highest levels since the genre’s early boom in 2023.

Let’s talk numbers: Downloads shot up by 67 per cent HoH in the first half of 2025, according to Sensor Tower, setting a new pace for growth. Meanwhile, consumer spending on IAPs doubled versus the previous six months.

People aren’t just downloading these apps—they’re getting hooked. Total time spent in Generative AI apps reached a mind-boggling 15.6 billion hours in the first half of 2025, which breaks down to more than 86 million hours every single day.

That kind of engagement led to 426 billion total sessions in just six months—translating to about 50 sessions for every person on the planet.

When ChatGPT first broke out, North America and other English-speaking regions raced ahead in adoption. Initially, North America made up about 20 per cent of global AI app downloads.

Fast forward to the first half of 2025, and North America’s share has dipped to 11 per cent, not because usage is shrinking there, but because world adoption—especially in Asia—has exploded.

AI Assistants steal the show

“Asia leads the way now, particularly in fast-growing markets such as India and Mainland China, where downloads surged 80 per cent from the previous half-year. Europe grew by 51 per cent, and North America still rose by a respectable 39 per cent,” the report showed.

Delving into the app types, the Generative AI market is largely split between AI Assistants and AI Content Generators. While content generators dominated until late 2022, the arrival of ChatGPT turned the spotlight to AI Assistants and chatbots.

Their momentum is overwhelming: by the second quarter of 2025, 85 per cent of all downloads in these categories came from AI Assistants like ChatGPT, Google Gemini, and DeepSeek.

Still, the distinction between assistant and content creation is blurring, as leading assistants roll out capabilities like image generation.

It has also become a competitive battleground. DeepSeek, launched in January 2025, made an immediate splash, outperforming all other Generative AI apps—including ChatGPT—for new global downloads in its first six months, thanks to its strong uptake in Asia, the Middle East, and Africa.

AI in non-game apps explode

ChatGPT remains the all-time download leader at 940 million, followed by Google Gemini’s 200 million and DeepSeek’s 127 million by June 2025.

Apps like Grok and Meta have joined the ranks of top performers, capitalizing on consumers’ growing appetite for intelligent chatbots.

Despite their global reach, these apps still attract a fairly specific crowd. Generative AI users skew younger, with a clear male bias—nearly 70 per cent of ChatGPT’s US audience is male, and 64 per cent are under 35.

Apps like DeepSeek, Claude, and Grok lean even more heavily male, while ChatGPT, Copilot, and Gemini show a bit more demographic balance (at least 30 per cent of users are women).

On the flip side, entertainment-centric apps such as PolyBuzz and Character AI are a hit with young women. ChatGPT users are also highly engaged, checking in about 13 times per month—similar to social favourites like X and Reddit, and even ahead of newer contenders like Threads.

But Google still holds the daily-use crown, with users logging in 18+ times a month.

A few years ago, from 2015 to 2019, AI was mostly a games phenomenon on mobile platforms. Since ChatGPT’s debut in 2022, the application of AI in non-game apps has exploded.

The software sector is leading the way, but Health & Wellness, Education, Lifestyle, Services, and Financial Services are quickly catching up: each saw over 200 new apps using AI-related terminology in the first half of 2025.

Subgenres like Photo Editing, Test Prep, Translation, Nutrition, Language Education, and Video Editing are especially popular places for AI integration.

The implications for the broader digital ecosystem are profound. ChatGPT is rapidly closing the gap with search engines and browsers as a daily go-to info hub.

In the first half of 2025, ChatGPT users averaged 7.8 sessions per day—a 37 per cent jump year-on-year—actually nudging past leading browser cohorts.

Although daily minutes spent is still a bit shy of Google’s, ChatGPT’s numbers are climbing fast, up 58 per cent over last year, hitting an average of 16 minutes daily.

The meteoric growth in sessions and usage signals a new digital habit for users and hints that AI Assistants like ChatGPT are steadily carving out a place alongside, or perhaps even ahead of, traditional search tools.

Mark Zuckerberg’s AI bets are already paying off

  • Meta’s core businesses—Facebook, Instagram, WhatsApp, and Messenger—collectively haul in $46.6b in ad revenue.

It seems like we’re living in the age of big ideas — and at Meta Platforms, Mark Zuckerberg is pushing the boldest one yet.

His promise? To deliver “personal superintelligence for everyone.” Meta’s latest financial moves reflect just how massive this ambition is.

During the most recent quarter, Meta spent a jaw-dropping $17 billion on capital expenditures, pouring most of that into boosting AI infrastructure and supercharging data centres. And this is hardly a one-off; the company expects these hefty investments to keep rolling in through at least 2026.

Despite these sky-high expenses, Meta’s shares surged 11.5 per cent after hours, celebrating some truly stellar second quarter results. Revenue soared 22 per cent to $47.5 billion, with Meta’s core businesses—Facebook, Instagram, WhatsApp, and Messenger—collectively hauling in $46.6 billion in ad revenue.

User engagement is stronger than ever, too: daily active users now total a stunning 3.5 billion people. And let’s not forget profit—net income rocketed 36 per cent to $18.3 billion over last year, showing that those big bets (so far) are paying off.

I keep thinking about how Zuckerberg defines Meta’s new mission: “The intersection of technology and culture is where Meta focuses.” He doubled down on this message in an Instagram Reel, underscoring that the new AI-focused lab is gearing up to build the next generation of advanced models.

Acquiring talent

What really catches the attention, though, is Meta’s approach to talent. Zuckerberg says he’s spent this quarter assembling an “elite, talent-dense team,” and he’s reportedly brought on board Alexandr Wang—one of the world’s youngest billionaires—to lead the charge.

Building the future takes more than ideas; it takes top-tier brains, dreamers, and doers. And apparently, Meta is becoming a magnet for them—because, as Zuckerberg puts it, Meta has “all of the ingredients required to build leading models and deliver them to billions of people.”

Here’s another big move: the new superintelligence team will get to play with “unparalleled compute” resources thanks to Meta’s investment in new gigawatt-plus computing clusters. That’s a geeky way of saying: a lot of power, ready to handle a lot of data.

But why sink so many billions into AI? Zuckerberg is crystal clear: “We’re making all these investments because we have conviction that superintelligence is going to improve every aspect of what we do.” If he’s right, Meta’s spending spree could transform not only its platforms but the entire way we interact with technology in daily life.

Weak AI chip sales and China curbs drag Samsung profit down 55%

  • South Korean chipmaker’s prolonged underperformance has triggered concerns among investors.

Samsung Electronics has faced a turbulent quarter, reporting a 55 per cent drop in second-quarter operating profit as key pressures continue to weigh on its semiconductor division.

For the April to June period, the company posted an operating profit of 4.7 trillion won (about $3.37 billion)—the weakest performance seen in six quarters and matching closely with their prior guidance of 4.6 trillion won. While revenue inched up 0.7 per cent to 74.6 trillion won, aligned with earlier estimates, these small gains haven’t done much to soothe investor nerves.

The heart of the challenge remains with Samsung’s chip division, traditionally its powerhouse. This quarter, the unit reported just 400 billion won in profit, a staggering fall from 6.5 trillion won the same period a year ago. This marks a notable milestone: it’s the first time in six quarters the division’s profit has dipped below the 1 trillion won threshold.

Samsung attributed this steep decline to several factors. The company cited adjustments to memory chip inventories and one-off costs related to US export controls, which limit the sale of advanced semiconductors to China, affecting the company’s contract chipmaking business.

Struggling to narrow technological gap

These headwinds are exacerbated by slow shipments of high-bandwidth memory (HBM) chips—an area where Samsung is struggling to narrow the technological gap with more nimble rivals supplying big players like Nvidia, especially for artificial intelligence data centre applications.

The chipmaker’s prolonged underperformance has triggered concerns among investors regarding its ability to catch up in the critical HBM market, an arena gaining importance with every leap forward in generative AI technologies.

There’s a sense of urgency now for the South Korean giant to innovate and regain commercial momentum.

On the bright side, Samsung secured a beacon of hope just days before these results: Tesla signed a $16.5 billion deal to source chips from Samsung, providing a potential lifeline for its foundry business, which manufactures chips on a contract basis.

The lucrative partnership with Tesla might help buoy the foundry division and offset recent losses, but the company still faces a challenging path ahead as it adapts to shifting global trade dynamics and escalating technological demands.

Microsoft gears up to spend over $100b in capex this fiscal year

  • For the current first quarter, the company predicts a jaw-dropping $30b in spending, breaking its own record for single-quarter investment.

If you ever wanted to see what major ambition looks like in the digital age, take a close look at Microsoft.

Marking its half-century milestone since Bill Gates and Paul Allen launched it in New Mexico back in 1975, the legendary tech company is today pouring the kind of money you can almost hear whistling through server racks on its way out the door.

These days, the numbers attached to Microsoft’s artificial intelligence dreams are almost as jaw-dropping as its stock price.

In the most recent fiscal fourth quarter, Microsoft once again bested Wall Street’s forecasts—a now-common achievement, as proven by its five-quarter streak of outperforming analysts.

Investors—maybe with confetti already in hand—watched as Microsoft’s shares hovered just shy of record highs, swelling by 22 per cent since the start of the year.

The company reported $76.4 billion in quarterly revenue, handily beating expectations and posting an 18 per cent year-over-year jump. That includes a sparkling $3.65 in earnings per share, well ahead of consensus.

So, where does all that cash go?

Into the great digital unknown, otherwise known as capital expenditures—and the amounts are staggering. Microsoft is gearing up to spend over $100 billion in capex in its next fiscal year, a 14 per cent jump from last year’s already-lofty tally.

For the current first quarter alone, the company predicts a jaw-dropping $30 billion in spending, breaking its own record for single-quarter investment.

If that figure turns out as forecasted, Microsoft could outpace fierce rivals Alphabet (Google) and Amazon in the race for world-dominating data center capacity over the next year.

This mad dash for digital horsepower isn’t just for bragging rights. Microsoft’s Azure cloud division is on a rocket ride, now surpassing $75 billion in yearly revenue thanks to the relentless demand for AI-powered services.

Satya Nadella, still steering the ship as CEO, said: “Cloud and AI is the driving force of business transformation across every industry and sector.”

Azure’s 34 per cent growth is proof positive that companies everywhere are shifting their computing needs to the cloud—and they want Microsoft’s brand of smarts to help them do it.

Big Tech’s capital spending

Of course, it’s not only Microsoft hurling money at the problem. Just last week, Google announced it would invest $85 billion in 2025—$10 billion more than it had planned—while Amazon is sprinting to meet its own $100 billion commitment.

This mass buildout means Big Tech’s capital spending might top $330 billion this year alone. Meta, not to be left behind, is also opening its wallet wide, but with a slightly cooler approach to capital outlays.

Money isn’t only flowing into hardware and buildings, either. The war for AI talent has gotten so feverish that $100 million signing bonuses are being dangled by Meta to lure experts from OpenAI, with reports of $200 million offers for top Apple engineers. Finding the brains who can actually build all this next-generation technology isn’t easy—or cheap.

Meanwhile, Microsoft isn’t shy about showing what all this investment is buying. Its Copilot AI now claims over 100 million monthly active users—a big milestone, though still dwarfed by Google’s Gemini, which boasts 450 million. As the numbers get bigger, so do the stakes.

Is it worth it? For now, investors certainly seem to think so. “I feel very good that the spend that we’re making is correlated to basically contracted, on-the-books business that we need to deliver,” CFO Amy Hood told analysts—her confidence mirrored by a $500 billion surge in AI stocks following Microsoft and Meta’s blockbuster quarters.

The game is on, and from the look of things, Microsoft is quite comfortable playing at these dizzying heights.

Russia blocks Ookla’s Speedtest to tighten internet control

  • Regulator suggestes that Russians switch to domestic alternatives amid push for digital sovereignty

Russia’s communications watchdog Roskomnadzor announced the nationwide blocking of Speedtest, the US-based internet performance tool operated by Ookla.

According to Roskomnadzor, the measure was taken to mitigate what it identifies as threats to the security and stability of Russia’s digital infrastructure, citing a 2020 decree aimed at reinforcing protections over the country’s segment of the internet.

Foreign technology companies, including major players like Alphabet, Meta, and Apple, have faced ongoing scrutiny from Russian regulators regarding the content they host and the locations where user data is stored.

The pressure on these companies has noticeably intensified over the past several years, especially in the wake of Russia’s invasion of Ukraine in February 2022, which accelerated the country’s efforts toward “digital sovereignty.”

Threaten user privacy

Roskomnadzor suggested that Russians switch to domestic alternatives, explicitly recommending ProSet—also known as ProNet in English—a homegrown application that offers internet speed and signal testing, similar to Speedtest.

The recommendation is part of a broader campaign to promote Russian-developed digital products and reduce reliance on Western technology.

This isn’t the first major move by the Kremlin to assert control over the national digital landscape. The Russian government has previously blocked major social platforms such as Instagram and Facebook (both owned by Meta), reduced the accessibility of YouTube, and pressed the messaging service WhatsApp to prepare for an exit from Russia.

A new messaging platform, MAX, developed under state oversight, is reportedly poised to take its place.

International organisations are watching Russia’s tightening grip with concern. In a report released, Human Rights Watch warned that the government’s increased surveillance capabilities and greater authority over internet infrastructure enable wider censorship and more comprehensive blocking of independent tools—even those designed to bypass restrictions.

“For years, Russian authorities have been carefully constructing a highly regulated and isolated segment of the internet,” observed Anastasiia Kruope, assistant researcher for Europe and Central Asia at Human Rights Watch.

New laws, she noted, significantly strengthen censorship and could threaten user privacy, with penalties now in place for online searches related to content the government labels as “extremist.”

The report further emphasised that these policies put Russian citizens at greater risk, as their personal data is more likely to be shared with law enforcement authorities under the new regulations.

Lenovo issues urgent BIOS updates for high-severity vulnerabilities

  • The vulnerabilities affect certain IdeaCentre AIO 3 and Yoga AIO all-in-one desktop computers.
  • While BIOS fixes for some Yoga AIO models aren’t expected until between September and November 2025, many others have updates already released.
  • The main risk: a privileged local attacker could exploit these flaws to read protected memory contents or even execute arbitrary code within System Management Mode.

Lenovo has addressed a set of critical BIOS vulnerabilities that put millions of devices at risk, releasing urgent firmware updates for multiple system models.

As of July 2025, while many updates are available, some models remain pending fixes, and Lenovo is actively communicating these risks to users.

The weaknesses were discovered in specific BIOS versions provided by Insyde Software, a major supplier of system firmware. According to Lenovo, these vulnerabilities affect certain IdeaCentre AIO 3 and Yoga AIO all-in-one desktop computers.

The main risk: a privileged local attacker could exploit these flaws to read protected memory contents or even execute arbitrary code within System Management Mode (SMM).

Six separate issues

SMM is a specialised operating mode running with exceptionally high privileges—beyond even the operating system (ring -2)—giving it the ability to halt virtually all other software on the system.

SMM is responsible for handling power management, direct hardware control, and various OEM-implemented low-level operations. As such, vulnerabilities in this area can have serious, far-reaching impact.

The vulnerabilities, first disclosed by the Binarly Research team in April 2025, comprise six separate issues: four rated as high severity (8.2/10 on the CVSS scale) and two rated as medium (6/10).

Successfully exploiting these flaws would let an attacker with existing local, high-level privileges escalate from kernel-level access (ring 0) to system management mode (ring -2), potentially exposing sensitive firmware memory (SMRAM) and allowing malicious code execution in this privileged context.

A powerful reminder

Such deep-level access could enable attackers to install persistent threats—malicious firmware capable of surviving a full wipe and OS reinstall, completely bypassing most conventional security defenses. However, exploiting these bugs is not trivial and would require the attacker to have pre-existing high-level (kernel) system access.

Lenovo’s official guidance is clear: users should check for the latest available BIOS updates for their affected systems and apply them as soon as possible.

While BIOS fixes for some Yoga AIO models aren’t expected until between September and November 2025, many others have updates already released. For maximum protection, keeping system firmware current is essential.

This situation is a powerful reminder of how critical firmware security—and timely patching—has become in today’s computing landscape. Staying up to date with vendor advisories and recommended security practices remains the surest way to protect against these evolving risks.