Palantir’s Meteoric Rise Surpasses Even NVIDIA — Who in China Can Match It?

Since 2023, Baidu’s Robin Li has been hammering home one core message: the potential of AI applications will ultimately surpass that of AI models themselves.

The rise of Palantir Technologies (PLTR) — famously known for its role in tracking down Osama bin Laden — is a vivid confirmation of that vision.

In 2025, Palantir has become Wall Street’s hottest AI stock. August was a breakthrough month: despite recent market fluctuations, the company remains at the center of investor attention, with daily trading volumes exceeding USD 10 billion and topping broker hot lists. Over the past year, its stock price has soared more than 400%, outpacing even giants like NVIDIA and Broadcom. The narrative around Palantir has shifted from computing power and large language models to something far more grounded — application-driven AI.

Palantir’s moat isn’t about the size of its models or the brute force of its compute. It lies in deep industry expertise. Rather than positioning itself as an omnipotent “silicon god,” Palantir acts more like a highly specialized industry consultant — an approach long overlooked by many Chinese tech firms, which tend to focus on building platforms and setting the rules rather than embedding themselves deeply into specific use cases.

Palantir’s story proves Li’s point: the imagination and value potential of AI in real-world applications may well be greater than creating AI itself.

01. Breaking the Conventional Wisdom

To understand who could become “China’s Palantir,” you first need to understand Palantir’s core DNA.

Traditionally, the AI companies making the most money have been platform players — NVIDIA with its compute platforms, or cloud giants like Azure and AWS — because they control the industry’s “rules of the game.”

Palantir challenges this model.

In August 2025, it delivered a stunning earnings report, with revenue surging 50%. Its market capitalization now places it among the top ten U.S. tech companies.

The first keyword that defines Palantir is application. Like Baidu, which emphasizes applied AI, Palantir is proudly dubbed “the first AI application stock.” Unlike Google, which started with a technology-first approach, Palantir began by embedding itself in practical use cases, only later adopting AI as a tool.

In simple terms: tech-centric AI companies make the hammer, then look for nails. Application-driven companies already know where the nails are — AI just makes the hammer better.

Palantir’s business is straightforward: Gotham for government and defense, and Foundry for enterprises. Its AI platform, Palantir AIP, is essentially an add-on to Foundry to lower adoption barriers — not a standalone new business.

The company focuses far less on model architecture and far more on how to deploy AI effectively. As its executives put it:

“LLMs are fragmented intelligence; ontology is the logic of true understanding. Without Palantir, LLMs cannot operate effectively in the real world.”

The second keyword is deep industry integration.

For example, during the pandemic, Palantir worked with the UK’s National Health Service to integrate patient data and resource availability into a unified model, boosting allocation efficiency by over 40%.

Most SaaS companies avoid customization because it reduces margins and scale benefits. Palantir embraces it — turning industry-specific expertise into a defensible moat. General-purpose models can be replicated with compute and talent; deep domain data must be earned over years.

Palantir’s top 20 clients spend an average of USD 70 million annually, making each new customer a major revenue event. This deep integration, coupled with technical strength, creates a lock-in effect that allows Palantir to avoid hyper-competition at the model layer — and enjoy extraordinary margins.

Its latest report shows an adjusted operating margin of 46%, net margin of 33%, and free cash flow margin of 57% — far exceeding Microsoft’s ~30%.

Even without further growth, Palantir is already an exceptional business. More importantly, it offers a strategic lesson for Chinese companies: application-focused AI may be just as valuable — if not more so — than general-purpose AI platforms.

02. Who Could Be “China’s Palantir”?

This is a point Robin Li has emphasized for years. In his view, the “model wars” are a misguided race. In the mobile internet era, super apps like WeChat and Douyin created value on par with, if not greater than, Android. In every tech wave, opportunities at the application layer tend to outnumber those at the system layer.

Following this logic, Baidu invests heavily in applications — which has translated into tangible results. In the first half of 2025, Baidu Smart Cloud won 48 contracts worth RMB 510 million, ranking first in both count and value.

Take smart transportation as an example, where both Palantir and Baidu have invested. Baidu’s “Highway Emergency Command AI Agent” has been deployed on the Jingxiong Expressway in Hebei, capable of detecting incidents like illegal parking in real time, generating rapid response plans, and achieving a 95%+ accuracy rate in early warnings.

Baidu’s approach — embedding AI deeply into specific industries and scenarios — mirrors Palantir’s philosophy. While Baidu also develops compute and model capabilities, these platforms exist to serve application delivery.

For instance, its Qianfan large model platform — now upgraded to version 4.0 — includes specialized models for finance, computer vision, and role-playing, enabling highly efficient and domain-specific applications. In gaming, NPCs in NetEase’s “Justice Online” interact with players in character-specific styles, powered by Qianfan.

Baidu also produces “out-of-the-box” AI agents, like its Yijian Visual AI platform, capable of building complex AI skills within minutes. One real-world deployment in Sany Energy’s factories alerts workers in real time to operational errors, significantly cutting inspection time.

Like Palantir, Baidu has a high-value client base — 65% of China’s central SOEs choose Baidu Smart Cloud.

Beyond big tech, China’s potential Palantirs could also be vertical technology service providers. These companies may not have cutting-edge AI capabilities but possess deep industry knowledge and proprietary data — from mapping (Baidu Maps, Amap) to logistics (JD Logistics, SF Express, Cainiao). Startups with years of niche data collection could also thrive, especially as platforms like Qianfan lower AI development barriers.

03. A Completely Different Valuation Logic

Despite operational similarities, Baidu and Palantir are valued under entirely different capital market logics.

Palantir trades at a rolling P/E near 500, the highest in the S&P 500 — a valuation NVIDIA itself doesn’t enjoy — reflecting Wall Street’s belief in its vast growth runway.

Baidu, by contrast, trades at just 8x rolling earnings, despite Q1 2025 Smart Cloud revenue growth of 42% and AI new business revenue topping RMB 10 billion with a 34% YoY increase.

Part of this gap comes from performance differences, but much is due to market sentiment. Chinese investors simply don’t buy into the “AI story” the way U.S. markets do. Even Alibaba — which has fully pivoted toward AI — trades at ~15x earnings and is still seen as a consumer company.

In 2025, Chinese capital markets appear far more enamored with consumer plays — the Hong Kong “Big Three” are not Horizon Robotics, Baidu, and Alibaba Cloud, but Pop Mart, Lao Feng Xiang, and Mixue Bingcheng.

There’s skepticism about AI’s ability to drive explosive revenue in China, partly because the domestic SaaS market is less mature and lacks a compute powerhouse like NVIDIA.

Yet Palantir’s rise offers a clear takeaway: deeply embedding AI into industry-specific problems, leveraging proprietary data, and building long-term expertise can unlock extraordinary growth — anywhere.

Whether it’s a tech titan like Baidu or a niche digital service provider, China has no shortage of contenders for the title of “China’s Palantir”. The ones who seize this opportunity will not just follow the AI wave — they’ll define it.