This week, Sinolink Securities released a tracking report on the technology industry. The data is limited, but when viewed together, it is quite interesting.

On one side, there are changes in the activity levels of AI chat assistants, and on the other, there are sales trends of domestic smartphones and PCs. These two industries, seemingly unrelated, actually point to the same issue: the previous round of technological competition is entering a phase of realization, and the data will reveal who has truly created something good.

01 Overseas AI Collectively 'Hits the Brakes,' Why Can Kimi Charge Ahead?

Let's start with AI.

The report includes a set of data on overseas chat assistants: ChatGPT saw a slight increase, Claude saw a slight increase, and Gemini saw a slight decrease.

Summed up in three words: remarkably stable.

This would have been unimaginable a year ago. Back then, every time ChatGPT released new features, its activity levels would surge. Now? New features are still being released, but user enthusiasm has clearly dulled. The reason is not complicated: the ceiling for conversational AI arrived earlier than expected. Writing copy, creating summaries, and modifying code—these high-frequency scenarios have already been covered, and the remaining low-frequency demands cannot sustain significant growth.

The real variable is in China.

The report mentions a sentence: 'Kimi's activity has surged due to the impact of new models.' No specific numbers are given, but this precisely indicates that the change is significant enough to be perceived without numbers.

Kimi's surge is noteworthy for two reasons.

First, it shows that domestic AI users have not formed an inertia of 'only using ChatGPT.' As long as your product experience is good enough, users are willing to switch. This is different from overseas—in the United States, ChatGPT's first-mover advantage is almost unshakable, and the switching cost is high. However, there is no absolute leader in the domestic market, and every model iteration is an opportunity to reshuffle the deck.

Second, the critical point where model capabilities transition from 'usable' to 'user-friendly' is being crossed. User activity does not surge simply because of a version with 'more parameters'; it must be a qualitative improvement in the experience of a specific scenario—perhaps the accuracy of long-text processing has improved, response speed has increased, or memory capacity has become stronger. Whatever it is, users are voting with their feet, and their votes are never based on technical specifications but on real experiences.

This is a positive signal for the entire domestic AI sector: the gap at the model layer is narrowing, and the reputation at the application layer is starting to play a real role.

02 Smartphone Market Warms Up by 1.3%, But the Real Story Isn't in This Number

Now, let's look at consumer electronics.

In June, domestic smartphone sales reached 21.48 million units, up 1.3% year-on-year. If you only look at the number, it's not impressive—1.3% is almost statistically negligible.

But against the backdrop of the past two years, this is a rare positive year-on-year growth. Since the replacement cycle was extended to nearly three years in 2024, the smartphone market has been struggling in a downward trend. The June figure at least suggests that the bottom may have been reached, and the market is looking for a position to stabilize.

In terms of market share, Huawei leads with 18%, followed by Apple at 15%, Honor and OPPO at 14% each, and Xiaomi at 12%. Compared to two years ago, a significant change is that Huawei is back.

This is not a 'nostalgic return' but a natural result of restored product momentum. After the gradual easing of chip supply issues, Huawei's channel capabilities and brand recognition in the mid-to-high-end market have hardly been substantially weakened. The Nova 16 ranked first in sales for a single model, indicating strong market appeal in the mid-range price segment.

Another noteworthy piece of information is that the iPhone 17 Pro and Pro Max both entered the top three. Apple's high-end dual-flagship strategy remains effective in China—the standard and Plus versions perform ordinary (average), but the Pro series has an extremely stable loyal user base unaffected by market fluctuations.

If you had to summarize the current state of the smartphone market in one sentence: the Head pattern (head pattern , meaning the competitive landscape among top players) is largely stable, and growth is driven by demand for high-end upgrades rather than new users. In other words, this is a market of inventory competition, where the key is who can make existing users more willing to spend money on upgrades, not who can acquire new users more aggressively.

03 Desktop Sales Up 11%, Laptop Sales Down 3%—What Does This Contrast Reveal?

There is a notable contrast in the PC data that deserves separate attention.

Laptop sales fell 3% year-on-year to 1.97 million units. Desktop sales rose 11% year-on-year to 1.5 million units.

Most people's intuition is that laptops are the future and desktops should have been phased out long ago. But the data tells us the opposite—in commercial and productivity scenarios, desktops are far more resilient than imagined.

The reason is not complicated. A friend who does AI training recently built a dual-GPU workstation. His exact words were: 'Running large models on a laptop is a joke—cooling and power consumption set the upper limit.' Such demands may not be mainstream, but they involve high unit prices and strong user loyalty. Moreover, as AI applications become more widespread, the population needing local high-performance computing power is expanding, not shrinking.

The slump in laptops, on the other hand, reflects more issues on the personal consumption side. Extended replacement cycles, increased price sensitivity, and declining appeal of new models—these are all familiar talking points, but no effective solutions have been found yet.

04 Three Uncertainties That Must Be Faced

The report concludes with three risk warnings. Many readers may habitually skip this section, but it is precisely the part most worth reading carefully.

The first is chip process nodes. Whether for AI chips or smartphone SoCs, improving yield rates for advanced process nodes is becoming increasingly difficult. This is not a problem specific to one company but a physical limit challenge for the entire industry. If the commercialization of nodes below 3nm is slower than expected, it will directly affect the performance and power consumption of end products, which are the things consumers care about most.

The second is policy. Sino-U.S. friction in the semiconductor sector has shifted from 'whether to impose restrictions' to 'what further restrictions can be imposed.' Every policy escalation affects not just a specific chip but the expectations of an entire supply chain. For the AI and smartphone industries, which rely on advanced process nodes, this is an external variable that cannot be ignored.

The third is consumer willingness. The aforementioned recovery is based on current economic expectations. If the macroeconomic environment changes, smartphones and PCs, as non-essential goods, will be the first categories consumers postpone purchasing.

Looking back at this report, although it is just a snapshot of data at a certain point in time, the structural information it reveals is worth following continuously.

The AI application layer is moving from 'everyone is about the same' toward differentiation. The fluctuations in activity levels driven by product iterations indicate that user loyalty has not yet been locked in, and a window of opportunity still exists.

The smartphone market has found temporary support in the bottom region, but the sustainability of the rebound depends on the synergy between product innovation and the macroeconomic environment. The trend has not yet been established, but signals have emerged.

The countertrend growth of desktops reminds us that some demands do not disappear just because they 'look old.' Technology advances, but the logic of usage scenarios is often more stable than the technology itself.

Data is always just a surface phenomenon; the real information lies in the comparisons and contrasts within the data.

Data Source: Sinolink Securities Research Institute, 'Kimi's Activity Surges, Smartphone Market Growth Warms Up in June'

Disclaimer: This article does not constitute any investment advice and is for reference only.