DeepSeek's $7B Round Makes It a Commercial Platform, and Its Brand Scores Already Favor Auto and Electronics in 2026

DeepSeek is shrinking in audience and growing in capital at the same time. QuestMobile counted about 127 million monthly users for the app in May 2026, down 24% year over year, while Doubao reached 368 million. Yet in the same month DeepSeek was reported to be raising roughly 50 billion yuan (about $7 billion) at a valuation of $52–59 billion. In hubGEO's tracking, DeepSeek also posts the highest average brand score of the six Chinese models we monitor (70.4 out of 100), and it scores international auto and electronics brands about four points higher than Doubao does. For international brands, that mismatch between reach and influence is the story of the second half of 2026.

What the funding round actually changes

According to The Next Web's report on the round, this is DeepSeek's first outside capital. Founder Liang Wenfeng is reportedly committing about 20 billion yuan of his own money, with Tencent considered for around 10 billion yuan and CATL around 5 billion yuan. TMTPost's June analysis puts Liang's control at 84.29% after the round, which means DeepSeek will not be run like a typical venture-backed startup. These figures come from people familiar with the matter and press reports, not from company filings, so treat the exact amounts as provisional.

Three consequences matter for brands. First, DeepSeek becomes a commercial entity with a board and investors, which pushes it toward revenue. Second, its backers are strategic: Tencent controls WeChat, which opened its AI interface to all mini-program developers on June 8, giving it a route into a network of about 1.4 billion monthly users. CATL is the world's largest EV battery maker and sits at the center of the auto supply chain. Third, the money buys compute, which is the real constraint in Chinese AI; TMTPost notes that DeepSeek's V4 model (open-sourced in April, 1.6 trillion parameters, 1 million token context) was cut to a quarter of its prior API price.

The practical reading is that DeepSeek's weight in brand discovery will increasingly be set by distribution partners and enterprise deployments, not by its standalone app. We covered the WeChat channel in our analysis of DeepSeek inside WeChat; the funding round makes that channel more durable.

The audience is smaller, the citation weight is not

Here is the scoreboard as of QuestMobile's H1 2026 ranking of AI-native apps.

Our average scores come from the 39 international brands currently in the hubGEO database, each queried 12 times per model. The sample is small and should be read as directional, not as a market census. Even so, the pattern is clear: audience size and brand score are only loosely related. Doubao has nearly three times DeepSeek's users and scores 1.7 points lower on average. Kimi's user base has collapsed by roughly half, yet its score is only five points below DeepSeek's, which tells us that scores reflect what a model knows, not how many people ask it.

For a marketer, that cuts both ways. A weak DeepSeek score is a less visible problem than a weak Doubao score because fewer consumers are asking. But DeepSeek answers also flow into enterprise tools, WeChat, and other products built on its open weights, and those surfaces are invisible to app-store ranking data.

Where DeepSeek diverges: auto and electronics

The more useful finding is category-specific. When we compare category averages across DeepSeek, Doubao and Qwen, DeepSeek's lead is concentrated in technical categories.

In auto, DeepSeek outscores Qwen by seven points. Tesla scores 78 on DeepSeek, 72 on Doubao and 65 on Qwen. Porsche goes 68, 62, 55. Dell, in electronics, scores 62 on DeepSeek against 52 on Doubao. In beauty the order flips: Qwen leads at 60.9, and SK-II shows the pattern clearly with 85 on Qwen, 78 on Doubao and 72 on DeepSeek.

Why would this happen? We cannot see inside any model, so this is an inference, not a measurement. DeepSeek is known for reasoning-heavy answers that reward spec sheets, technical comparisons and structured explanation. Auto and electronics are categories where those documents exist in abundance. Qwen's strength in beauty is consistent with Alibaba's retail data: Tmall reviews and product listings are exactly the material that beauty shoppers search for. Food and luxury, where brand recognition is near-universal, show almost no gap between models. KFC, for example, scores 94 on DeepSeek and 95 on Doubao.

The gaps are also not uniform within a category. Lexus (52 on DeepSeek, 42 on Qwen) and Volvo (42 versus 40) sit in the lower half of the auto tier on every model, which shows that DeepSeek's edge lifts brands that already have some footprint but does not rescue brands that are absent. We saw the same dynamic in our auto brand visibility analysis.

Why the CATL and Tencent money points the same direction

The investor list reinforces the category tilt without proving it. CATL's presence signals that DeepSeek sees industrial and automotive customers as a strategic market. Tencent's presence points to consumer distribution through WeChat. If both strategies work, DeepSeek will be answering more questions about vehicles, batteries and devices, and more questions from people who are already inside a chat app and will not open a separate search window.

That matters because DeepSeek's V4 launch showed how fast the model's behavior can change. As we noted in our DeepSeek V4 analysis, a new model version can move brand scores in either direction within weeks. A commercial DeepSeek with paying enterprise customers will ship updates on a more regular cadence, not less.

There is a countervailing risk. Investors may push DeepSeek toward monetization features, including sponsored or priority placement, as Doubao has begun doing with paid tiers and advertising. DeepSeek has not announced anything of the kind, and we are not predicting it. But brands that rely on organic citation today should assume the organic window is not permanent, and should build their evidence base while it is open.

How to read a score gap in practice

A four-point category gap does not mean a four-point swing in consumer behavior. What it signals is that the model has more, or more consistent, material about your brand in that category. Read it as a relative signal of how consistently a model surfaces your brand across the queries we run. The spread inside a single model is far larger than the spread between models: on DeepSeek alone, Porsche scores 68 and Volvo 42, a 26-point gap between two brands in the same category. Small average gaps between models therefore hide large differences in how individual brands fare, and that is where a funded DeepSeek's category focus could move real outcomes. The practical test is simple: ask each model an open recommendation question in your category, in Mandarin, and note whether your brand is volunteered or only appears once you name it.

What the data does not tell us

Two limits are worth stating plainly. The funding figures are press-reported and may change when the round closes. And our brand sample is 39 brands with 12 queries per model, so a four-point category gap could narrow with a larger sample. Treat the numbers as a prompt for your own testing, not as a ranking you can quote to a board.

Takeaway for Brand Marketers

1. Test DeepSeek separately if you sell technical products. If you are in auto, electronics, or other spec-driven categories, DeepSeek is the model where well-structured technical content is most likely to pay off. Audit how it describes your models, trims, and specifications, and correct errors at the source.
2. Do not over-weight DeepSeek if you sell beauty. Qwen and Doubao carry more of the category's discovery, and Qwen's Alibaba-linked sources (Tmall reviews, product pages) matter more there than on DeepSeek.
3. Publish reasoning-friendly content. Comparison tables, test results, and clear specifications are the formats DeepSeek's answers appear to favor. Put them on authoritative Chinese-language pages, not only on your global site.
4. Watch the WeChat surface. DeepSeek's reach inside Tencent's ecosystem will not show up in app MAU rankings. Include WeChat-based queries in your monitoring.
5. Re-baseline after the round closes. A funded, commercial DeepSeek is likely to release more often. Schedule a quarterly check of your brand's score on all six models rather than a single annual audit.

Related: Compare your brand's scores across all six Chinese models on the hubGEO brands page, and read Doubao's paywall economics for the other side of the monetization story.

Sources: The Next Web, TMTPost, QuestMobile via Tencent News, hubGEO brand score database.