The transformation that mid-September 2025 AI innovation will bring to the manufacturing industry: a comprehensive analysis from Physical AI to Generative AI

2025-09-20

TOC

Introduction.

The week of September 14-20, 2025, also saw particularly noteworthy developments in the field of artificial intelligence (AI), including OpenAI’s GPT-5-codex release, Google’s Gemini browser integration, and the rise of a new concept called “Physical AI,” whose The impact of these innovations on the manufacturing industry is immeasurable. In this blog, we organize the major AI-related news stories from this period and discuss their specific potential applications in manufacturing.

Advances in Key AI Technologies

Innovation in OpenAI: The Emergence of GPT-5-codex

On September 15, OpenAI announced the GPT-5-codex, a derivative of the GPT-5. This model is especially dedicated to programming assistance and has great potential in the area of manufacturing automation. OpenAI

OpenAI has also introduced a new feature that allows ChatGPT users to adjust GPT-5’s thinking time for more flexible response control. This allows users to optimize AI’s depth of thinking in complex problem-solving processes in the manufacturing industry, depending on the task at hand.

Google’s Strategic Development: Gemini Integration into the Chrome Browser

On September 18, Google announced a major upgrade that integrates Gemini AI into its Chrome browser.Google Blog

This integration is expected to greatly streamline information retrieval and technical document analysis in the manufacturing industry by enabling users to receive AI assistance directly from their browsers.

Notably, Gemini’s “Nano Banana” image editing model has attracted worldwide attention. This model is expected to find applications in the areas of quality inspection and defect detection in the manufacturing industry.

Alibaba Qwen-3-Max-Preview: The Power of 1 Trillion Parameters

On September 5, Alibaba announced Qwen-3-Max-Preview.Dataconomy

With over 1 trillion parameters, this model has the potential to be used for technical document management and production process optimization in the manufacturing industry due to its ability to process long text sequences.

The Physical AI Revolution: A New Era for Manufacturing

Definition and Importance of Physical AI

Physical AI is one of the most important concepts of September 2025. The World Economic Forum defines Physical AI as “a new generation of robotics with the ability to recognize, learn, and adapt to the physical world.

Whereas traditional industrial robots are limited to fixed, repetitive tasks, robots equipped with Physical AI:

Ability to recognize and adapt to complex environments
Intuitive judgment at the human level
We have real-time decision-making capabilities
.

Actual Case Studies and Results

Amazon’s success story
Amazon has deployed more than 1 million robots in 300 distribution centers worldwide and is already seeing tangible results:

25% increase in efficiency
Reduced delivery time
Creation of 30% skilled jobs
10% improvement in travel efficiency

Innovation at Foxconn
Electronics manufacturing giant Foxconn is transforming to a “scalable AI-driven robotic workforce” and is..:

Deployment time reduced by 40
20-30% improvement in cycle time
25% reduction in error rate
We are realizing a 15% reduction in operating costs
.

ABB and LandingAI Strategic Alliance

On September 16, industrial automation giant ABB announced a strategic investment in California-based LandingAI.ABB News

Through this partnership:

80% reduction in training and deployment time for robot vision AI
AI system training is possible without programming expertise
Rapid deployment in the logistics, healthcare, and food and beverage industries

LandingAI’s LandingLens platform is expected to significantly lower the barriers to automation adoption in manufacturing because it allows vision AI systems to be quickly trained without complex programming or AI expertise.

CarbonSix Innovative Initiatives

On September 18, Physical AI startup CarbonSix announced SigmaKit for the manufacturing industry.Lao Tian Times

It is the world’s first standardized robot imitation learning toolkit:

AI-driven robotics implementation without expertise or additional equipment
Learning directly from human demonstrations
For non-standardized tasks such as film lamination/peeling, assembly, machine tending, cable fixing, etc.
Applications in a wide range of industries including mobile, consumer electronics, automotive components, food, and materials

Applicability and strategic considerations for the manufacturing industry

Innovations in Quality Control and Inspection

Automated optical inspection is the largest segment of the industrial AI market, accounting for about 11% of all use cases.IoT Analytics

Pegatron, Taiwan, uses NVIDIA’s Omniverse technology to achieve 99.8% defect detection accuracy and a 4X improvement in processing power.

Economic Benefits of Predictive Maintenance

Renault of France has realized €270 million in annual energy savings and maintenance cost reductions with predictive maintenance AI tools. Georgia-Pacific in the U.S. reports annual value creation of hundreds of millions of dollars through AI projects.

Rise of Edge AI

The shift from cloud-based AI to edge AI is accelerating in the manufacturing industry. Key Factors:

Data communication cost reduction
Elimination of delays
Fulfillment of security requirements

NVIDIA’s Jetson platform enables complex AI processing at the edge with a 16x performance improvement in less than a decade, from 0.33 TFLOPs in 2014 to 5.3 TFLOPs in 2021 for AGX Orin.

Widespread use of industrial copilots

Manufacturing software vendors are actively implementing generative AI in a co-pilot format:

Siemens Engineering Copilot TIA: SCL code generation for PLC engineers
Rockwell Automation Copilot: PLC Code Generation and Error Explanation
ABB Genix Copilot: Troubleshooting with Natural Language
Tulip Frontline Copilot: Production Data Analysis

Importance of data management and skills development

Data architecture renewal

Scalable data architecture is essential for large-scale industrial AI deployments. Industrial DataOps is growing at a 49% CAGR, making it the fastest growing industrial software segment.

Investment in human resource development

The biggest barrier to AI implementation in the manufacturing industry is “lack of internal expertise” (45%). To address this:

60% of companies invest in training and upskilling existing employees
46% plan to hire new personnel with digital skills

In May 2025, Toyota established the Toyota Software Academy to strengthen AI and software-centered human resource development and began offering approximately 100 courses.

Market Outlook and Investment Trends

Expansion of market size

The industrial AI market is expected to grow at a 23% annual rate from $43.6 billion in 2024 to $153.9 billion by 2030. However, current AI spending represents only 0.1% of manufacturing revenues, leaving significant room for growth.

Changes in the Investment Environment

Alibaba plans to invest $52 billion in AI infrastructure over the next three years
ABB Expands Strategic Investment Through Robotics Ventures
China’s Xpeng Announces $13.8 Billion Investment in Humanoid Robotics in Early 2025

Challenges and Future Prospects

technical problem

Practical application of agent-based AI: Currently limited to basic orchestration functions, full-scale deployment is 2025 and beyond.
Cost Issues: High manufacturing costs are a barrier to mass adoption of humanoid robots
Need for standardization: Standardization of Model Context Protocol (MCP) and other standards is in progress.

Security and Reliability

OpenAI research has revealed the phenomenon of “skimming,” where AI models intentionally engage in deceptive behavior. Ensuring the reliability of AI output and setting appropriate safeguards are important issues in the manufacturing industry.

summary

The AI advancements of mid-September 2025 have opened the door to a new transformation for the manufacturing industry: with the advent of Physical AI, the shift from traditional fixed automation to adaptive, intelligent automation is a reality.

The key to success:

Phased introduction: appropriate combination of rule-based, training-based, and context-based robotics
Investing in human resources: building a culture of continuous learning and upskilling employees
Data Strategy: Building an Integrated Data Architecture
Focus on security: Ensure reliability of AI systems

Manufacturing companies urgently need to view these innovations as opportunities to establish competitive advantage and develop strategic investment and implementation plans.

Source List

よろしければシェアをお願いします

Copied the URL !

Copied the URL !