FEAT
ATURE
ARTIFICIAL INTELLIGENCE
BEING SMART WITH AI Will Intellectual Property prove crucial once again?
Words by Partner, Peter Gray and Technical Assistant, Szymon Pancewicz of Intellectual Property Firm, Mathys & Squire
Despite setbacks in recent years, it appears that the AI market is starting to realise its potential, particularly in relation to Industry 4.0 and the embedded market. However, there’s a risk of losing originality within the marketplace. Intellectual Property (IP) specialist talks about how to compete in this lucrative sector
W
hile artificial intelligence (AI) and machine learning (ML ) are
seeping into an ever more diverse range of applications, it is hardware design that is currently predict ed to provide the foundations for t he next big breakthrough. The race is on to est ablish the fundamental AI accelerator technolo gy that will support the AI ecosystem, and as the market mat ures, intellectual property (IP) will prove critical to its success.
The growing ML model accura cy ove r recent y ears has been largely driven by increases in the size of ML models. Model size has roughly doubled around every 3.5 months. To maintain t his Moore’s- law-style growth, there is a clear need for new t ypes of AI accelerators that can support more complex ML models for both training and inference. The global AI chip market wa s
valued at $6.64 billion in 2018, and is projected to grow substantially in upcoming y ears, reaching $91.19 billion by 2025. Understandably, a wide range of companies are working to develop AI chips. However, the market is poised to go through a growth cycle similar to those seen in the CPU, GPU an d baseband processor markets: to be dominat ed by a few large players. The range of companies competing in the AI chips market spans both chip giants and niche ent ities. The race remains open as to who may dominat e; both technical specialists and investors will be looking closely at which companies’
technolo gy shows most promise , and t he field will inevitably evolve t hrough invest ment, acquisitions and failures until the market leaders emerge.
10 APRIL 2020 | ELECTRONICS
The current front runners appear to be Intel and Nvidia. According to R euters, Intel’s processors currently dominate the market for AI inference, while Nvidi a dominat es the AI training chip market. Neither Intel nor Nvidia are resting on their laurels, a s showcased by Intel’ s acquisition of Habana Labs for $2 billion, and Nvidia’s new Jet son Xavier NX edge AI chip. Several smaller entities also look exciting, such as Mythic and Graphcore, the latter havi ng recently partnere d with Microsoft, raising $150 million at a $1.95 billion valuation. At t his stage, it is unclear who will eventually dominate the AI chips market. And yet, a key lesson from historical development s, such as in the fields of CPUs and baseband processors, is that IP rights play an important role in who comes out on t op, and who survives in the long run. A number of market leaders in the chip industry have built their business models around pat ent licensing. Notable examples include Qualcomm and Arm, who derive the bulk of their profits from pat ent licensing, which will be equally profitable for companies who build a strong patent portfolio in the AI chips arena. Arm’s business model will present an attractive opt ion for startups t hat lack the resources to get involved in chip manufacturing; t he incentive to mitigate risk, by remaining fabless, will cont inue as well, even as smaller companies expand in stat ure.
“At this stage, it is unclear who will eventually dominate the AI chips market. And yet, a key lesson from historical developments, such as in the fields of the CPUs and baseband processors, is that IP rights play an important role in who comes out on top...”
“At this stage, it is unclear who will eventually ly
For startups that may be interested in being acquired, there is little doubt that IP is vital for the st rongest valuations. It is unlikely that Int el would have acquired Habana if it weren’t for the company’s patent portfolio, s tretching back to 2016. Investor exit strategies, therefore, continue to dictate t he need for a sound IP strategy.
Another key lesson from related
sectors is the very large risk and reward associated with pat ent infringement. As recent ly as January 2020, Apple and Broadcom were ordered to pay $1.1 billion in damages for infringing Cal Tech patents on Wi-Fi t echnology: the need for companies to build patent portfolios, for both offensive and defensive purposes, remains clear.
IP issues have not been missed by companies in the AI chips race. There are currently over 2,000 pat ent families in the field of AI chips, with Intel having filed 160 patent applications in this field just in the last five y ears. Thus, existing market leaders and new entrants should be mindful not to miss the importance of IP, and aim to protect t heir technology as early as possible.
Although the legal environment surrounding IP - pat ents in part icular - are ever changing, and the patent landscape is becoming more crowded, there is little doub t that IP will onc e again prove crucial in this emerging industry . Observers can look forward to a new wave of patent battles, and seasoned technologist s and IP
practit ioners can refine their approaches using the lessons of the past.
Mathys & Squire
www.mathys-squire.com / ELECTRONICS
Despite setbacks in recent years, it appears that the AI market is starting to realise its potential, particularly in relation to Industry 4.0 and the embedded market. However, there’s a risk of losing originality within the marketplace. Intellectual Property (IP) specialist firm Mathys & Squire talks about how to compete in this lucrative sector
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36
