Having handed in their "exam papers" on the same day, have both Baidu and Xiaomi reached their inflection points?

Baidu's "Fake Fall", Xiaomi's "Real Pain". 

Both have seen a sharp drop in profits, but their pain points are different. On August 18, Baidu and Xiaomi released their financial reports for the second quarter of 2026 on the same day. Baidu's net profit fell 68% year-on-year, while Xiaomi's adjusted net profit dropped 42.6% year-on-year. 

Judging only from these two sets of data, the two industry giants are like two comrades in distress whose profits have been nearly halved. The feedback from the market is also very realistic: Baidu's stock price fell after hours, and Xiaomi was also widely viewed with pessimism. 

However, if you break down the two financial statements, they are completely different scenarios.

Of Baidu's reduced profits, about 4.6 billion yuan came from the fair value withdrawal of investment assets and exchange rate effects. This is a floating loss recorded on the books, not money flowing out of operations, and will not hurt the foundation of its business operations. 

If this part of the profit and loss is excluded, Baidu's operating profit is about 3 billion yuan, and its operating profit margin remains at around 10%. 

Almost all of Xiaomi's declining profits come from its business side: skyrocketing storage costs have eaten into the gross profit of its mobile phone business; the reduction of IoT subsidies has dragged down sales revenue; car deliveries increased by 28.2% year-on-year, but the business is still loss-making; innovative businesses such as AI cannot yet support large-scale revenue. 

With the three layers of cost pressure piling up, Xiaomi can only earn less, and every penny of lost profit is a real loss of flesh and blood. 

In other words, Baidu's book fluctuation amplifies the real operating pressure, while Xiaomi has a real chunk of its profit statement eroded.

Many views hold that the worst period for Baidu and Xiaomi is passing, and the inflection point of secondary growth is already in sight. 

But to verify whether the inflection point is real, we need to go back to the financial reports to answer a judgment: who is "faking a fall" and exchanging profits for the future? Who is experiencing cyclical pains without replacing its growth engine? 

Xiaomi's Real Pain: Half Cycle, Half Trade-off

The most impressive figure in Xiaomi's financial report is the mobile phone ASP: it rose 25.9% year-on-year to 1351 yuan, hitting a record high; the proportion of models priced above 3000 yuan reached 32.1%, up 4.5 percentage points year-on-year. 

The cost behind this is that the shipment volume of smartphones dropped from 42.4 million units to 31.2 million units, down 26.5% year-on-year; the gross profit margin also fell from 11.5% to 8.5%. Beyond the financial statements, Xiaomi also has to withstand external "sentiments" such as doubts from shareholders, pressure from the supply chain on order reduction, and market concerns about market share. 

However, in the more than four years since Xiaomi started its high-end push, it is very necessary to make active trade-offs. It needs to calculate its own accounts for what to sell and what not to sell, and can no longer rely on luck to make profits. 

According to Lu Weibing's statement at the conference call, "we will no longer pursue mobile phone shipment volume in the short term." 

This shows that facing cost pressures such as rising memory prices, Xiaomi has chosen to actively reduce the shipment of low-end mobile phones and prioritize ensuring profits. 

There is another set of data that is easy to be "glossed over". In the second quarter, the gross profit of internet services was about 6.9 billion yuan, surpassing the gross profit of about 6.3 billion yuan from IoT for the first time, becoming the business with the highest gross profit contribution for Xiaomi. Among them, advertising revenue was 7.2 billion yuan, up 4.8% year-on-year, with a gross profit margin of 76.8%, up 1.4 percentage points year-on-year. 

When both the mobile phone and IoT businesses are under pressure, this high gross profit business acts as the profit stabilizer for Xiaomi. Its value has been underestimated by the market.

In the past, Xiaomi's profits relied on hardware businesses such as mobile phones, but now internet services rank first in gross profit contribution. The ecological value of the "Human-Vehicle-Home" system has propped up revenue when hardware income is under pressure. 

Then look at two large investments: automobiles and AI. Automobile deliveries reached 104,000 units, up 28.2% year-on-year; the loss was 2.6 billion yuan, narrowing from the 3.1 billion yuan loss in the first quarter. For the automobile business, narrowing losses is more convincing than revenue growth. 

This means that the scale effect of Xiaomi's automobile business has begun to appear, and costs are being diluted. At the same time, the Pengcheng extended-range SUV is pre-sold at a price range of 260,000 to 300,000 yuan, targeting family users, with low user overlap with the SU7, which perfectly fills the space for the vehicle model structure and gross profit margin. 

Sedans build brand value, SUVs drive sales volume, and walking on two legs is the way to stability. 

Xiaomi's approach to AI is also very Xiaomi-style: it does not rush to make profits independently, but first embeds AI into devices to nourish the ecosystem. AI revenue related to the MiMo large model is included in "other related businesses", which totaled 1 billion yuan in the quarter, and API calls and Token solutions have begun to generate revenue. 

Although the revenue scale is not disclosed separately, the commercialization path is already clear. The financial report specifically mentions that the test success rate of humanoid robots in its own factory has exceeded 90% and will be publicly displayed at the World Robot Conference, which is Xiaomi's way of telling the market that it has more cards in hand. 

In terms of valuation, Xiaomi's current price-earnings ratio is between 15 and 17 times, with net cash of about 176 billion yuan, equivalent to about HK$10.5 per share; it has repurchased 11.7 billion Hong Kong dollars of shares within the year, and has launched a new share repurchase plan of 20 billion Hong Kong dollars. 

However, the market is still pricing Xiaomi according to the standards for hardware companies. If the gross profit margin of the automobile business continues to improve, MiMo revenue is disclosed separately, and overseas IoT sales grow rapidly, this pricing standard will have to change. The market is offering a hardware valuation, while the company is offering a confidence valuation through share repurchases, and the gap between the two prices is the expectation difference in valuation logic. 

But the inflection point is not up to Xiaomi alone. As analyzed by Zimobang, whether it can deliver on expectations depends on three variables: can Pengcheng break through in the cooling extended-range vehicle market? Can the actual experience of the Xuanjie O3 meet expectations? Can its AI business continue to stay in the SOTA camp? 

Baidu's "Fake Fall": Two Sets of Accounts in Conflict, the Inflection Point Has Arrived

Baidu's accounts need to be read in reverse. 

Online marketing, that is, advertising revenue, was 13.1 billion yuan, down 19% year-on-year. This is Baidu's traditional growth engine, which is indeed decelerating. However, AI cloud revenue reached 7.3 billion yuan, up 50% year-on-year; GPU cloud has maintained triple-digit growth for four consecutive quarters, with a growth rate of 283% this quarter. 

Fitch downgraded Baidu's rating, but at the same time predicted that AI business revenue will exceed traditional business revenue in 2026. This shows that the institution recognizes the growth speed of Baidu's new growth engine, but given that it has not yet seen profit realization, it gives a cautious evaluation. 

This is the conflict between the old and new sets of accounts, which leads to seemingly contradictory judgments from the outside world.

One detail is worth noting: JPMorgan Chase gives Kunlunxin an independent valuation of 40 billion to 49 billion US dollars, of which the portion belonging to Baidu is about 27 billion to 34 billion US dollars. The market has begun to split Baidu from "one company" into "multiple assets" for separate pricing. 

This is a substantial revaluation of its technical reserves. 

Smart money is also moving. According to 13F position documents cited by NetEase Technology, Druckenmiller established a new position in Baidu ADR in the second quarter, which is his first position in a major Chinese concept stock since he liquidated his Alibaba position in 2023; Tepper also increased his holdings in the same period. 

Baidu's inflection point has already appeared in its business operations. In the second quarter, the revenue of Baidu's core AI-driven business reached 12.5 billion yuan, up 25% year-on-year, accounting for half of the general business revenue; in the first quarter of this year, this proportion was 52%. The proportion has decreased slightly, but the 25% year-on-year growth rate is real, and the new growth engine has not decelerated. 

Baidu's incremental business lies in vertical industries and Robotaxi. The number of benchmark cases of AI cloud in fields such as finance, gaming, and embodied intelligence continues to expand, with higher stickiness of vertical solutions and stronger bargaining power. Apollo Go has been operating in 28 cities, with a total test mileage of 350 million kilometers. It has realized fully unmanned commercial operation in Dubai, reached cooperation with Uber, and public road tests in London, Switzerland and other regions have also been launched one after another. 

Once the overseas model is replicated, its valuation will switch from a "domestic mobility service provider" to a "global autonomous driving technology exporter". 

NetEase Technology observed that the market's questions have changed: from "when will AI generate revenue" to "whether this capability can be replicated". From office agents to GPU cloud, from self-developed chips to Robotaxi, the same path is being repeatedly proven: technology turns into products, and products turn into customers. 

Of course, Baidu is also going through pains. For example, the gross profit margin dropped from 43.9% to 39%, and its business has become more asset-heavy. Both AI cloud and GPU cloud are businesses that require heavy investment in computing power before generating returns, and the decline in gross profit margin is normal during the investment period. The revenue of the AI application layer was 2.5 billion yuan, with the growth rate slowing to 3%, and C-end monetization still takes time. 

But fortunately, Baidu's core business foundation is stable, cost control is effective, and the transformation is out of control. Its future direction is much clearer than it was a year ago. 

In my opinion, Baidu's current transformation is like an Odyssean journey home: Ithaca has not disappeared, but on the way home it must pass through the sea of the Sirens. Advertising revenue is that tempting song, but the mast has been tied, and the direction has not changed. 

Its old revenue decreases by a part, and new revenue makes up for it, and the speed of supplementation determines the degree of pain. The proportion of AI revenue exceeding half, the independent valuation of Kunlunxin, and the landing of overseas Robotaxi, these cards have already been placed on the table. 

For Li Yanhong, there is no turning back. The sunk investments, established relationships, and formed revenue structure are all pushing Baidu forward. 

Two Types of Pain, Different Inflection Points

Looking at the two financial reports together, the inflection points of the two companies have different meanings. 

Xiaomi's inflection point lies in healing external wounds. Storage costs, the reduction of IoT subsidies, and automobile losses are all blows to its business. These external wounds can all "heal on their own": for example, storage prices follow a cycle, the reduction of subsidies is a one-time event, and automobile losses are narrowing. 

What Lei Jun needs to do is to stabilize the morale of the team, and continue to move forward in the direction of building a high-end, full-stack AI ecosystem company. 

But before that, Xiaomi still relies heavily on the help of "industry cycles" to wait for the rebalancing of supply and demand. Its underlying growth engine has not been completely replaced. When the next industry cycle comes, what will its response strategy be? 

Baidu's inflection point means that the switch of its business growth engine is more than half completed. The AI business's ability to generate profits independently is quickly filling the gap left by the old business. Cloud, API, and Token solutions all need to generate profits on their own, which is also forcing it to find its own profit logic faster. 

What Li Yanhong needs to prove is whether the new AI business can make profits, and whether the planned commercial blueprint can be implemented strongly. 

The capital market is very snobbish, and it is used to defining a company by its profits. A drop in share price means "it's failing", and a rise means "it's doing well again". 

But money never lies. Xiaomi's capital flows to the trade-offs for high-end development, overseas IoT, automobile scale expansion and AI ecosystem, waiting for the cost cycle to fall and new businesses to grow; Baidu's capital flows to AI infrastructure, Kunlunxin, vertical industries and Robotaxi, waiting for new revenue to fill the gap of old revenue. 

It is not easy for the two companies that are at their respective inflection points: one is waiting for the cycle to turn back, the other is waiting for the engine to be fully replaced, and this is the real highlight worth paying attention to. 

This article is from the WeChat Official Account "Tang Chen's Notes", author: Tang Chen, published with authorization from 36Kr.