Patent wars occur in areas where money is concentrated. Since 2010, smartphone technology has begun to develop rapidly with the mass distribution of smartphones, and the smartphone market has never been quiet due to patent litigation as various technologies are adopted in order to be chosen by consumers. Numerous lawsuits have been filed against smartphones, including semiconductors, software, operating systems, interfaces, games, camera technology, and battery technology. Korean companies that used to produce smartphones such as Samsung Electronics, LG Electronics, and Pantech naturally began to be exposed to patent litigation as global smartphone sales increased, and they began to be busy responding to patent litigation in the United States, Europe, and Japan. In the case of my patent attorney exam classmates, I used to get a job as an in-house patent attorney for companies such as Samsung, LG, Hynix, and Hyundai Motor Company, and spearhead litigation with domestic and foreign patent firms.
In the late 2010s, as most smartphone technologies became more common, patent wars became less frequent than before. Although the patent specialized company (NPE) business, which does not manufacture but only secures patents, is still doing well in the United States, most of the patent disputes related to smartphones have been resolved, and smartphones have become a market in which 'cost competition' technology is more dominant than 'advanced technology'. Even if they are not manufactured in Korea, smartphones with high-cost competitiveness using low labor costs in China, Vietnam, India, and Indonesia can perform well, and the market is changing as low-cost smartphones have become reliable in terms of quality. The third industrial revolution represented by the Internet and mobile and the tumultuous patent wars that took place are gradually losing momentum.
Now, the Fourth Industrial Revolution, represented by 'generative artificial intelligence', is clearly beginning. I thought that since the big framework of information processing had not changed anyway, there was no need to distinguish between the 3rd and 4th industrial revolutions. However, transformer models such as GPT are certainly new technologies that are different from the IT technologies of previous generations, and generative AI will change our lives tremendously, so we had no choice but to accept the concept of the Fourth Industrial Revolution. Recently, as the generative AI company leaderboard above shows, many generative AI companies have appeared. Money is pouring into the field of generative AI, and it is becoming clear that patent wars will be fought in this field in the future.
Artificial intelligence can be divided into hardware and software, and technological development and market expansion are rapidly being achieved due to their compatibility. GPU processors, represented by Nvidia, are fueling the growth of generative AI software companies such as OpenAI. In addition, with the emergence of numerous generative AI software companies, data centers are constantly being created to handle the computations they need. Naturally, AI hardware such as NVIDIA continues to work on the servers located in the data centers, and the competition among Big Tech companies is accelerating because companies with better GPUs will have AI hegemony in the future. This is similar to the mid-2010s, when smartphone hardware, smartphone OS, and smartphone APP continued to develop each other, promoting the development of communication technology and data centers.
Naturally, artificial intelligence-related patents are increasing rapidly. Patents that protect artificial intelligence technology are divided into hardware patents and software patents. AI hardware patents are emerging in an infinite number of fields, such as AI semiconductor design and production technologies such as GPUs and NPUs, memory structures of AI semiconductors such as HBM, communication methods between semiconductors, and methods of supplying power to this hardware. NVIDIA has already built a strong patent portfolio in the United States, China, Germany, Taiwan and South Korea, as shown in the table below. The number of patent applications has been increasing every year, and it can be seen that it has increased even more recently. Like Intel and Qualcomm in the past, NVIDIA will build strong intellectual property barriers with patents, and hardware AI patent portfolio barriers will strongly check latecomers.
Nvidia is already involved in 71 patent lawsuits. Now they are mainly in the position of being compensated for damages. This is the case in the patent litigation market, but if there is a rumor that a person is making a lot of money, there will be more lawsuits in that direction. For example, patent companies (NPEs) purchase patents that have been filed and registered by universities or research institutes a few years ago and make a profit by pursuing patent litigation against large corporations such as NVIDIA. The fact that we are starting to see the news that NPEs have started their activities means that there is a high possibility that patent wars will be active in the artificial intelligence hardware market in the future.
Hynix and Samsung Electronics, which make HBM, a memory optimized for AI processing, and their suppliers who supply materials, parts, and equipment to them will also have to prepare in advance for patent wars that will soon occur in the field of artificial intelligence semiconductors and memory. In the current situation, where the U.S. government is encouraging production in the U.S., it is essential to prepare for patent disputes that will arise in the U.S., as consumption, parts, and equipment made by Korean partners will be used in U.S. factories.
AI software patents can be developed in a more diverse way. Currently, there are contents related to CNNs, RNNs, and transformers, which are related to the basic algorithms of artificial intelligence, and more patents related to various neural network structures will appear in the future. These algorithms are being registered as patents, and they will grow into important patents that will receive large compensation in the next five years.
Among AI software patents, patents related to data processing methods are becoming particularly important. The recent 'TPU (tensor processing unit) patent litigation between Google and Singular Computing' clearly shows the importance of AI patents. Singular Computing (the plaintiff) filed a patent lawsuit in 2019, claiming that Google (the defendant) used it to develop TPU without permission, even though it shared its technology through several meetings with Google from 2010 to 2014. TPU is Google's dedicated hardware designed to accelerate machine learning and AI tasks, and it plays a key role in AI technology.
Google filed a trial to invalidate Cingular Computing's U.S. patent No. 10,416,961, but failed, and Cingular Computing's patent survived. Singular Computing claimed that it had suffered up to $7 billion (about 9.24 trillion KRW) in damages due to Google's patent infringement, and although the settlement amount was not known, the two sides finally reached an agreement and ended the dispute in January 2024. This case is a case in which patents related to data processing methods played an important role, and it can be said that it will be an important indicator in the future patent wars for artificial intelligence software.
While preparing a recent lecture plan, I searched for Kipris, a patent information search service, and found that there were more than 1,000 AI-related patents represented by BLT, the patent firm I am working with. Among AI software patents, patents that deal with neural network structure and data processing methods are good, but if you are a startup that is actually aiming for sales, I think the most realistic strategy is to focus on service patents that apply AI. Since matters related to the neural network structure and data processing methods of generative AI are far from generating immediate sales, I think it is an area that should be intensively researched by research institutes, universities, and large corporations that are supported by R&D research funding. Since it is not very expensive to come up with an idea and experiment with it and turn it into a patent, I think it is okay to create an aggressive patent portfolio if you have a new idea in this field.
On the other hand, as in the example below, artificial intelligence companies such as 'Seven Point One' and 'AiOfarm' are startup companies that provide services by applying artificial intelligence technology to reality. Seven Point One has a solution that exchanges questions with potential dementia patients through artificial intelligence and analyzes the interlocutor's answers to diagnose the possibility of dementia. Most local governments in Gyeonggi Province are participating, and they are preparing to enter the United States in recognition of their high efficiency.
AiOfarm has a solution that automatically selects fruits based on artificial intelligence. It is a company that has technology that can quickly screen not only apples, tangerines, and melons produced in rural areas, but also agricultural products with various shapes such as mushrooms, ginseng, and strawberries with an artificial intelligence camera system to quickly identify top-grade agricultural products and fruits with defective insides.
It is similar to the proliferation of smartphones in the early 2010s, when numerous companies that created apps that successfully identified people's needs continued to grow. In the future, there will be a huge number of companies related to AI application services in the field of artificial intelligence in Korea. Since OpenAI, Google, and Microsoft cannot target all industries, companies that apply artificial intelligence to the market through detailed observation and service planning will be successful. They should build a patent portfolio by appealing to what kind of artificial intelligence they use and how their business model will have any effect. This will be the first way to seize business opportunities in the AI patent war that we will soon encounter.
A patent is a type of insurance in which the state promises a certain period of exclusivity to the inventor who proposes a new configuration. In the future, numerous patent lawsuits will occur in the field of artificial intelligence, which attracts money, and companies that make weapons in advance will have the upper hand.
By Uhm Jeonghan
Patent Attorney, BLT Patent & law Firm www.BLT.kr
Investment company BLT & partners www.bnp.ac
Patent wars occur in areas where money is concentrated. Since 2010, smartphone technology has begun to develop rapidly with the mass distribution of smartphones, and the smartphone market has never been quiet due to patent litigation as various technologies are adopted in order to be chosen by consumers. Numerous lawsuits have been filed against smartphones, including semiconductors, software, operating systems, interfaces, games, camera technology, and battery technology. Korean companies that used to produce smartphones such as Samsung Electronics, LG Electronics, and Pantech naturally began to be exposed to patent litigation as global smartphone sales increased, and they began to be busy responding to patent litigation in the United States, Europe, and Japan. In the case of my patent attorney exam classmates, I used to get a job as an in-house patent attorney for companies such as Samsung, LG, Hynix, and Hyundai Motor Company, and spearhead litigation with domestic and foreign patent firms.
In the late 2010s, as most smartphone technologies became more common, patent wars became less frequent than before. Although the patent specialized company (NPE) business, which does not manufacture but only secures patents, is still doing well in the United States, most of the patent disputes related to smartphones have been resolved, and smartphones have become a market in which 'cost competition' technology is more dominant than 'advanced technology'. Even if they are not manufactured in Korea, smartphones with high-cost competitiveness using low labor costs in China, Vietnam, India, and Indonesia can perform well, and the market is changing as low-cost smartphones have become reliable in terms of quality. The third industrial revolution represented by the Internet and mobile and the tumultuous patent wars that took place are gradually losing momentum.
Now, the Fourth Industrial Revolution, represented by 'generative artificial intelligence', is clearly beginning. I thought that since the big framework of information processing had not changed anyway, there was no need to distinguish between the 3rd and 4th industrial revolutions. However, transformer models such as GPT are certainly new technologies that are different from the IT technologies of previous generations, and generative AI will change our lives tremendously, so we had no choice but to accept the concept of the Fourth Industrial Revolution. Recently, as the generative AI company leaderboard above shows, many generative AI companies have appeared. Money is pouring into the field of generative AI, and it is becoming clear that patent wars will be fought in this field in the future.
Artificial intelligence can be divided into hardware and software, and technological development and market expansion are rapidly being achieved due to their compatibility. GPU processors, represented by Nvidia, are fueling the growth of generative AI software companies such as OpenAI. In addition, with the emergence of numerous generative AI software companies, data centers are constantly being created to handle the computations they need. Naturally, AI hardware such as NVIDIA continues to work on the servers located in the data centers, and the competition among Big Tech companies is accelerating because companies with better GPUs will have AI hegemony in the future. This is similar to the mid-2010s, when smartphone hardware, smartphone OS, and smartphone APP continued to develop each other, promoting the development of communication technology and data centers.
Naturally, artificial intelligence-related patents are increasing rapidly. Patents that protect artificial intelligence technology are divided into hardware patents and software patents. AI hardware patents are emerging in an infinite number of fields, such as AI semiconductor design and production technologies such as GPUs and NPUs, memory structures of AI semiconductors such as HBM, communication methods between semiconductors, and methods of supplying power to this hardware. NVIDIA has already built a strong patent portfolio in the United States, China, Germany, Taiwan and South Korea, as shown in the table below. The number of patent applications has been increasing every year, and it can be seen that it has increased even more recently. Like Intel and Qualcomm in the past, NVIDIA will build strong intellectual property barriers with patents, and hardware AI patent portfolio barriers will strongly check latecomers.
Nvidia is already involved in 71 patent lawsuits. Now they are mainly in the position of being compensated for damages. This is the case in the patent litigation market, but if there is a rumor that a person is making a lot of money, there will be more lawsuits in that direction. For example, patent companies (NPEs) purchase patents that have been filed and registered by universities or research institutes a few years ago and make a profit by pursuing patent litigation against large corporations such as NVIDIA. The fact that we are starting to see the news that NPEs have started their activities means that there is a high possibility that patent wars will be active in the artificial intelligence hardware market in the future.
Hynix and Samsung Electronics, which make HBM, a memory optimized for AI processing, and their suppliers who supply materials, parts, and equipment to them will also have to prepare in advance for patent wars that will soon occur in the field of artificial intelligence semiconductors and memory. In the current situation, where the U.S. government is encouraging production in the U.S., it is essential to prepare for patent disputes that will arise in the U.S., as consumption, parts, and equipment made by Korean partners will be used in U.S. factories.
AI software patents can be developed in a more diverse way. Currently, there are contents related to CNNs, RNNs, and transformers, which are related to the basic algorithms of artificial intelligence, and more patents related to various neural network structures will appear in the future. These algorithms are being registered as patents, and they will grow into important patents that will receive large compensation in the next five years.
Among AI software patents, patents related to data processing methods are becoming particularly important. The recent 'TPU (tensor processing unit) patent litigation between Google and Singular Computing' clearly shows the importance of AI patents. Singular Computing (the plaintiff) filed a patent lawsuit in 2019, claiming that Google (the defendant) used it to develop TPU without permission, even though it shared its technology through several meetings with Google from 2010 to 2014. TPU is Google's dedicated hardware designed to accelerate machine learning and AI tasks, and it plays a key role in AI technology.
Google filed a trial to invalidate Cingular Computing's U.S. patent No. 10,416,961, but failed, and Cingular Computing's patent survived. Singular Computing claimed that it had suffered up to $7 billion (about 9.24 trillion KRW) in damages due to Google's patent infringement, and although the settlement amount was not known, the two sides finally reached an agreement and ended the dispute in January 2024. This case is a case in which patents related to data processing methods played an important role, and it can be said that it will be an important indicator in the future patent wars for artificial intelligence software.
While preparing a recent lecture plan, I searched for Kipris, a patent information search service, and found that there were more than 1,000 AI-related patents represented by BLT, the patent firm I am working with. Among AI software patents, patents that deal with neural network structure and data processing methods are good, but if you are a startup that is actually aiming for sales, I think the most realistic strategy is to focus on service patents that apply AI. Since matters related to the neural network structure and data processing methods of generative AI are far from generating immediate sales, I think it is an area that should be intensively researched by research institutes, universities, and large corporations that are supported by R&D research funding. Since it is not very expensive to come up with an idea and experiment with it and turn it into a patent, I think it is okay to create an aggressive patent portfolio if you have a new idea in this field.
On the other hand, as in the example below, artificial intelligence companies such as 'Seven Point One' and 'AiOfarm' are startup companies that provide services by applying artificial intelligence technology to reality. Seven Point One has a solution that exchanges questions with potential dementia patients through artificial intelligence and analyzes the interlocutor's answers to diagnose the possibility of dementia. Most local governments in Gyeonggi Province are participating, and they are preparing to enter the United States in recognition of their high efficiency.
AiOfarm has a solution that automatically selects fruits based on artificial intelligence. It is a company that has technology that can quickly screen not only apples, tangerines, and melons produced in rural areas, but also agricultural products with various shapes such as mushrooms, ginseng, and strawberries with an artificial intelligence camera system to quickly identify top-grade agricultural products and fruits with defective insides.
It is similar to the proliferation of smartphones in the early 2010s, when numerous companies that created apps that successfully identified people's needs continued to grow. In the future, there will be a huge number of companies related to AI application services in the field of artificial intelligence in Korea. Since OpenAI, Google, and Microsoft cannot target all industries, companies that apply artificial intelligence to the market through detailed observation and service planning will be successful. They should build a patent portfolio by appealing to what kind of artificial intelligence they use and how their business model will have any effect. This will be the first way to seize business opportunities in the AI patent war that we will soon encounter.
A patent is a type of insurance in which the state promises a certain period of exclusivity to the inventor who proposes a new configuration. In the future, numerous patent lawsuits will occur in the field of artificial intelligence, which attracts money, and companies that make weapons in advance will have the upper hand.
By Uhm Jeonghan
Patent Attorney, BLT Patent & law Firm www.BLT.kr
Investment company BLT & partners www.bnp.ac