Will there be blockbuster AI applications by 2025? The answer seems to indicate a 50% chance.

This conclusion comes after consulting numerous industry practitioners, including entrepreneurs focused on AI, investors, and mid-level management from major internet companies engaged in AI developmentCompared to the beginning of 2024, these experts have become noticeably more cautious regarding their expectations for AI applications in 2025.

At first glance, this prediction might appear pessimistic, implying there is a significant probability that standout AI applications may not materializeThe surge in generative AI that began around the end of 2022 has now spanned over two rigorous years; hence, what could prevent the emergence of a major player in the market by 2025?

One AI entrepreneur, who once envisioned an explosion of consumer-facing AI products by the latter half of 2024, has reversed this sentiment

He has decided to focus on refining his products instead of hastily investing in expansive marketing efforts.

His rationale is straightforward: the current state of foundational technology, consumer mindset, and the overall ecosystem is not conducive to the emergence of a blockbuster AI application just yetAchieving a breakthrough in AI applications by 2025 will still rely on a degree of luck; without it, the odds of hitting the jackpot diminish significantly.

Zhang Yu, a leading venture capitalist and founding partner at Qingzhi Capital, has observed the evolution of investments in the AI sector from massive foundational models to specialized industry applications, and now to applied AIHe forecasts that by the end of 2024, there may finally be some hope for an explosive rise in AI applicationsHowever, he estimates the likelihood of achieving AI applications with tens of millions of daily active users by 2025 is low; the presence of billion-user blockbuster apps might require another two to three years.

Historically, the emergence of blockbuster applications in both the internet and mobile internet eras has universally depended upon time and chance.

The commercial use of the internet began in the mid-1990s, with the global industry undergoing a significant bubble burst by 2000. Companies such as Microsoft, Google, and Yahoo emerged in the West, while in China, platforms like Sohu, Sina, NetEase, and Tencent rose to prominence

Fast forward to the mobile internet era, and you find companies like Meituan, Didi, and Toutiao established around 2010, yet they only gained widespread visibility in the public consciousness a few years later, becoming household names after significant competitionNotably, short-video platforms like Douyin and Kuaishou did not become standout products until after 2017.

It’s entirely possible that transformative AI applications are already on the market, waiting to be discovered by the masses.

Currently, the majority of AI applications are still in their embryonic stagesAccording to statistics from Exceptional Capital’s research arm, as of the end of October 2024, there are only around 2,000 AI products globally that achieve monthly traffic exceeding 10,000 visits.

The popular AI applications of the moment are still considerably distant from attaining blockbuster status

Within the mobile internet sphere, metrics like “monthly active users” and “average daily usage duration” serve as key indicators of success, often with apps reaching billions of monthly users and exceeding two hours of daily usage.

Take for instance Doubao, a conversational AI app launched by ByteDance, which currently holds the title of the most-used AI application in China with 59.98 million monthly active users reported as of December 2024. However, the average daily usage duration on Doubao stands at just 6.1 minutesThis brief engagement time is echoed across various other conversational AI applications, which max out around seven minutes of daily usage.

Even entertainment-focused AI applications that theoretically should attract users for longer periods, such as Xingye and Maoxiang in China, fail to significantly extend engagement times

alefox

According to AI product rankings, Xingye's average daily usage is about 6.24 minutes, while Maoxiang's is a mere 2.68 minutesThere remains a scarcity of profitable avenues for these applications, as evidenced by Character AI, which was sold to Google after grappling with high training costs and limited paying subscribers, posing challenges for profitability.

Despite the challenging environment, professionals in the field remain optimistic about the futureThey are confident that an explosion of AI applications is a certain trend and that substantial AI-powered solutions will eventually ariseBy 2025, domestic technology supporting foundational AI models should be able to catch up with international developments and meet growing user demandsThe competitive pricing wars among major companies will drive costs down to mere cents per thousand tokens, enabling several applications to gain traction.

Zhang Yu noted that by the latter portion of 2024, he witnessed an impressive wave of AI applications advancing rapidly, with some even doubling their revenue within just a month or two