07 Industry Application Layer (B2B)

1. Industry Chain Positioning and Value Distribution

This is the ultimate arena where AI technology delivers commercial value.

In the "Smiling Curve," the industry application layer should naturally occupy the high-profit zone on the right end, but the current reality is "much thunder, little rain." Due to large model hallucination issues, many traditional SaaS products are merely AI chatbots bolted onto existing shells (wrapper products), failing to touch core enterprise workflows—resulting in extremely low customer willingness to pay. The real value lies in going deep into vertical industries and using proprietary data to build an irreplaceable intelligent decision-making core.

2. Core Application Scenarios and Breakthrough Challenges

Current B2B adoption shows a clear "easy-first, hard-later" pattern:

• Easy-to-deploy tracks (high fault tolerance): Intelligent customer service, marketing copy generation, low-code/no-code development generation (Copilot), legal contract initial review. These scenarios fall under "efficiency tools"—even if AI makes mistakes, humans can step in as a safety net.
• Deep-water tracks (extremely low fault tolerance, high barriers):
  • Autonomous driving (end-to-end large models): A full shift from early rule-driven approaches to data-driven end-to-end models; Tesla FSD V12 is the trendsetter in this direction.
  • AI + drug discovery (AI4Science): Protein folding prediction (AlphaFold), target discovery—dramatically compressing multi-billion-dollar drug R&D cycles.
  • Defense and high-end manufacturing: e.g., Palantir's intelligence analysis systems, which demand extremely high security and highly complex ontology-based data integration.

3. Going Global and Localization of Domestic Applications

The domestic enterprise services (ToB) market already suffers from weak willingness to pay. Combined with the high cost of AI-driven transformation, pure-play AI SaaS startups in China face severe survival pressure. Therefore, "application going global"—leveraging China's highly efficient engineering talent dividend to earn US dollars—has become the hottest trend.

---

4. Showcase of Core Benchmark Companies in This Layer

Humane, Inc.

Private | United States

Wearable AI Devices

Core Business: Developing AI-powered personal wearable interaction devices designed to replace smartphones

⚔️ Moat: Unique laser projection interaction technology and AI Agent platform, with deep software-hardware integration creating a first-mover advantage

Key Competitors: Apple, Meta (Reality Labs), Rabbit Inc.

Salesforce

CRM:US | United States

AI CRM and Enterprise Software

Core Business: The world's largest customer relationship management (CRM) software provider, fully integrating Einstein GPT / Copilot features to embed AI across the entire enterprise sales, service, and marketing workflows.

⚔️ Moat: Monopolizes the core sales and customer data of a vast number of global enterprises, with AI capabilities directly embedded in business workflows rather than standalone add-on tools, resulting in extremely high customer stickiness.

Key Competitors:

Tesla

TSLA:US | United States

Autonomous Driving (End-to-End Large Models) / Humanoid Robots

Core Business: AI-driven electric vehicles, Full Self-Driving (FSD), and commercial solutions for humanoid robots

⚔️ Moat: The world's largest real-world driving data loop + end-to-end AI model (FSD V12) + vertically integrated manufacturing capability, building dual moats in autonomous driving and robotics

Key Competitors: Waymo, Cruise, BYD, XPeng, UBTech

iFlytek

002230:SZ | China

AI+Education/Healthcare/Government

Core Business: Leveraging the Spark Cognitive Large Model, it provides industry solutions such as AI education, AI healthcare, and AI government-enterprise services. It is one of the few domestic companies to achieve AI commercialization through integrated software and hardware.

⚔️ Moat: Has extremely deep G-end (government) and B-end channel barriers in key domestic industries such as education and healthcare, backed by years of accumulated industry data and scenario understanding. The software-hardware integrated product forms (learning machines, medical auxiliary terminals) provide clear payment vehicles for AI capabilities.

Key Competitors:

CloudWalk Technology Co., Ltd.

688327:SH | China

AI Solutions and Project Outsourcing

Core Business: Provides customized AI solutions for finance, security, transportation and other industries, and undertakes enterprise-level AI project outsourcing

⚔️ Moat: Core technology platform based on human-machine collaborative operating system, but gross margin under pressure in project-based model, reliant on downstream customer relationships

Key Competitors: SenseTime, Megvii, Hikvision, iFlytek