2021 AI Security Keywords of the Year: Difficult

There are all kinds of cries and laughs, and it is difficult to say the head.

”

Author | Gao Xiusong

Edit | Yu fast

If you want to use one word to summarize the AI security circle in 2021, "internal and external difficulties" is slightly appropriate.

From the perspective of the general trend, the epidemic has been happening for two years, and there is no sign of termination in the short term: almost all major foreign economies have fallen, local epidemics have occurred in China from time to time, and both major markets are facing severe economic situations.

Externally, enterprises to go outward encountered more resistance, on the one hand, the purchasing power of customers in overseas markets declined, and enterprises had to sell products preferentially, exemplified by the year-on-year decline in gross profit margins; on the other hand, the United States has pulled a number of head companies into the investment blacklist, and the external situation is not optimistic.

Internally, the AI investment boom is gradually receding, the growth rate is slowing down, the financing Matthew effect is intensifying, and the AI companies that have grown up have taken away most of the financing; coupled with the FACT that AI itself has entered the landing stage, the fragmented scenario has increased the difficulty of landing, and it has not yet run out of a suitable commercialization path, and many AI companies are holding their wallets to live.

Difficulty is the most real feeling of almost all practitioners. From the "survival" of Hu Yangzhong, president of Hikvision, from the "survival" of Hikvision President to the "must intersect with other industries" at the Artificial Intelligence Security Summit at the end of the year, all of which corroborate the fact that this industry is facing an inflection point of growth.

However, the times are both pressure and motivation. Beyond the boundaries of difficulty, there are also some new changes. For example, the emergence of new growth poles, the potential energy of the scene with urban AIoT as the core gradually exerts force; for example, the market is further optimized, and the requirements for the technology itself are refined; for example, when the tide of capital retreats, the naked swimmers appear and the explorers stand at the head of the tide...

Under the pulse of the times, AI Nuggets has accompanied the industry through another year, witnessed the integration and mutual promotion of security and AI, and provided the industry with the most cutting-edge market change information, technology trends, new opportunities and new perspectives from an objective and neutral perspective.

Nowadays, looking back at the past year, AI Nuggets has summarized the important "hot search" of this year, and found the changes and opportunities of "security + AI" from these hot search keywords to entertain readers.

1

Keyword one: out

Market competition is cruel, some enterprises enter, and some enterprises withdraw.

For the outgoing, the farewell is always inevitable, but it is silent.

The defeat of 360 visual technology is a microcosm of the failure of many AI vision companies to nugget the security market. On this track, there are hai daewoo in the front and four small dragons in the back, and the entrants either follow the big brother to eat vegetables and drink soup, or dig a new market in a unique way.

360 visual technology undoubtedly wants to be a lone traveler: with visual security as the core, relying on the advantages of 360 Group in security, to cut into the AI security market.

"Safety, that's the foundation of everything." Qiu Zhaoqiang, then CEO of 360 City Intelligent Group and general manager of 360 Vision, said in an interview with AI Nuggets that 360 should "do the safest face recognition in the industry and be the first in the face recognition market within 2-3 years." ”

Unfortunately, this grand goal had only just set sail, and the ship's sails were blown off by a strong wind.

In July 2021, AI Nuggets learned that Qiu Zhaoqiang, general manager of 360 Vision, had left his job. The sales team of 360 vision research and development products has been basically cut, and employees have either transferred or left, "the entire business sector has been greatly adjusted." ”

AI Nugget Zhi cross-confirmed and got a positive reply, at the end of the year's Security Expo, there is no longer a 360 visual figure, but as a "billion square cloud" (data service) to participate, from the side to support the above news.

Many practitioners believe that this gale originated from the data security protection law.

In 2020, the draft data security protection law was officially released, and almost all enterprises have realized the security problem of data security, and have made efforts in this regard, from chips to terminals, from the edge of the cloud, all emphasizing the word security.

There is a threshold for safety, but for an old tycoon like Hai Daewoo and a technical rookie like the Four Little Dragons, this threshold does not seem to be enough to form a moat with absolute superiority.

When 360 Vision takes security as the core of the "sword to the peak" to cut into the market, but the business has not yet been rolled out, did not occupy enough market, other manufacturers turned back to the product to do security, kill a carbine, based on its mature business network, it seems that it is not difficult to squeeze out the former.

The above practitioners told AI Nuggets that in the case of market saturation, the business has no increment and is not worth 360's continued investment, so it is cut, which is also reasonable.

At the same time as the 360 visual exit, ping an technology is unknown.

In the past two years, Ping An Technology's personnel changes have been comparable to earthquakes, and CEO Chen Liming, CTO and chief architect Fang Guowei, Chief Data Officer Weng Zhi, and Chief Information Officer Pang Jing have left one after another.

It is hard to imagine that Ping An Technology, which once went out of the circle as a smart city, has gradually drifted away in the security circle after several years of prosperity.

"Transforming to digital consulting." An industry insider close to Ping An told AI Nuggets, "The executives engaged in technology have gone almost the same." ”

After the departure of former CEO Chen Liming, who was based on technology, Chen Xinying became the actual controller of Ping An Technology, which had worked at McKinsey for 13 years, was responsible for business technology consulting services, and was leading the transformation of Ping An Technology into a consulting company.

"Under the wave of digitalization, it may be easier to win by transforming into consulting." The industry insider said that after Chen Liming left, Ping An Technology's emphasis on consulting should only increase.

At present, Ping An Technology prefers to say "internal empowerment" and has no ambition to conquer security and AI.

AI Nugget believes that in the context of industrial digitalization, the transformation of Ping An Technology to a consulting company may be a way out, but whether it can be done has yet to be tested. (For more insider information about Ping An Technology, please refer to: Ping An Technology has once again left two "chief" executives, and a new "business strategy" has surfaced))

Another layer of this transformation means that the AI security circle is becoming more and more difficult to mix.

When AI security walks into the deep water area, it is bound to have higher requirements for the "swimming skills" of enterprises, and without excellent technology, it is difficult to "flow this mixed water".

The example of 360 and Ping An Technology is an example, which does not mean that they do not have the strength to do technology and products, but when considering the development of business, enterprises must consider the cost and benefit problem from the market pattern and development trend, when the income is unclear and the cost expenditure is predictable, the transformation of other businesses is normal.

But no matter what, the temporary fading of the two companies is also announcing the difficulties and sufferings of this industry, opening the prelude to the "withdrawal of the circle" of enterprises.

2

Keyword two: IPO listing

At the end of 2021, the biggest news in the AI security circle is senseTime's successful landing in Hong Kong stocks.

SenseTime's road to listing was slightly bumpy, and after submitting the prospectus, it was suddenly sanctioned by the United States when it was about to go public, but finally rang the bell on the Hong Kong Stock Exchange on the penultimate day of 2021.

The birth of the first AI stock means that the AI industry is gradually bidding farewell to adolescence and moving towards maturity, reflecting the recognition of AI by capital, but from a long-term perspective, whether it can tell ai stories well is a huge test. (For more details on SenseTime's listing, see: SenseTime's Revelations: A History of "Management Art" and "Technological Evolution"))

It is worth noting that SenseTime is not the only one favored by capital, Megvii, Yuncong, Yuntian Lifei, Geling Deep Pupil, etc. have successfully passed the meeting and are preparing for the final sprint.

However, in the view of AI Nuggets, listing can provide AI companies with more diversified financing methods to expand their business and develop technology, but they still have not run out of a suitable business model, the chronic disease of hematopoietic difficulties still exists, and the pressure on revenue will be greater than before.

"AI companies can only prove themselves with positive net profits, but they still can't see the boundaries."

In addition to the news of the listing of AI companies that have attracted much attention, AI Nuggets also noticed that security giant Hikvision is also undergoing a "slimming" campaign, intending to split the fluorite network and Hikvision robots for listing.

Usually, the spin-off and listing of subsidiaries is not only conducive to the independent development of subsidiaries, such as financing, talent incentives, management, etc., but also supports the diversified development strategy of the parent company. After the split of the subsidiary, the self-financing and no longer rely on the parent company for blood transfusion, the enthusiasm of employees can be improved, and the competitiveness of the brand is also a test.

Hikvision will be Fluorite network and Hikvisor robot split listing action, indicating its in-depth layout in the field of smart home and robots, after the split, Hikvision will take video capabilities as the core, expand the business line to industry, energy, transportation and other scenarios, nugget intelligent Internet of Things.

It is undeniable that capital operation is indeed an important driving force for the development of enterprises.

But fundamentally speaking, capital operation is a means rather than a way of survival for enterprises. Only by doing products and services in a down-to-earth manner can enterprises form a complete business closed loop. Otherwise, the castles in the air, which are piled up with capital operation, will eventually collapse.

3

Keyword three: data network

In 2021, from the perspective of policy direction, the implementation of the Personal Information Protection Law and the Data Security Law will play a role in setting the tone for the industry.

This means that enterprises must ensure data security in the process of data collection, transmission, storage, analysis, etc.

The two laws form a giant web that regulates enterprises in this industrial chain.

The foundation of AI is data, and it is only by relying on data to train algorithms. On the front-end perception side, companies should consider whether they are compliant when collecting data, and the diversity of their data sources and data types will be reduced; on the data transmission and storage side, each link of the cloud-side terminal core needs to encrypt the data to prevent leakage, which requires the technical strength of the enterprise itself; on the analysis side, sensitive data or special data are not suitable for algorithm training.

For example, in the past, enterprises could collect face data and do business innovation, but now they can only use faces in certain special scenarios, and their use is limited.

"The era of data security compliance has arrived." Xie Can, a senior researcher at Tencent Security Yunding Lab, said that the Personal Information Protection Law and the Data Security Law have two characteristics: one is a wide range, including domestic, cross-border and different industries, there are corresponding legal compliance requirements; second, the punishment is large, and the fine imposed by the Personal Information Protection Law on illegal enterprises is 5% of the annual turnover.

However, while regulating enterprises, Zhang Data Network has also spawned a demand for data security protection, and data companies that have labeled network security and privacy computing are ushering in a period of outlet. (For more data security, refer to: The Data Security Law is officially implemented today, should companies "panic" or "hug"?) ）

"On the surface, the cost of enterprises has increased to a certain extent, but the entire industrial chain is actually more perfect, which is beneficial in the long run."

AI Nugget Believes that the introduction of the Personal Information Protection Law and the Data Security Law is both a trend and a practical need.

In the early stage of the industry, when the government guides the development of a certain industry, it pays more attention to the role of the industry in promoting economic development; and when the industry enters a mature stage, its focus will inevitably tend to be people-centered services, and the compliance requirements of the industry will become a necessity.

The industry must be within the scope of the law to go further.

4

Keyword four: core change

Chip problems have a long history.

As early as last year, AI Nuggets noticed changes in the pattern of security chips, see: HiSilicon "out of stock", security "missing core".

From 2019 onwards, the US sanctions are pressing forward step by step, the living space of HiSilicon has been further narrowed, the shipment volume has been greatly discounted, and in the field of back-end chips, from the previous market share of 60%-70% to about 30% (data source: Everbright Securities), the share of front-end chips has also declined sharply.

Correspondingly, domestic chip companies such as Fuhan Micro, Guoke Micro, Beijing Junzheng, Zhongxing Micro, and Xingchen Technology have risen rapidly, and foreign companies such as Ambar and Texas Instruments have also made a comeback, aiming to regain the high-end chip market.

The data shows that in 2020, in terms of IPC Soc chips, HiSilicon's market share fell to 30%, and Xingchen Technology (25%), Fuhan Micro (17%), Beijing Junzheng (12%) and Guoke Micro (5%) divided 60% of the market.

It can be seen that in the field of front-end chips, the market share given up by HiSilicon's fall is effectively undertaken by domestic enterprises, and it can basically achieve localization.

However, in terms of DVR and NVR Soc chips, although domestic companies have accelerated their layout and research and development efforts, the main markets are occupied by foreign companies such as Ambarah and TI, and about 60% to 70% of HiSilicon customers have to turn to foreign companies with high prices and strict cooperation conditions.

In the field of high-end chips, after HiSilicon, there is still no company that can carry the beam.

An interesting example is that at this year's security expo, Huawei has been absent, but HiSilicon still exists. ON THE DAY OF THE AI NUGGET, DUE TO A SLOW STEP, THE PROMOTIONAL MATERIALS AND BUSINESS CARDS ON THE HISION BOOTH HAVE LONG BEEN "OUT OF STOCK", AND THE CROWD ON THE BOOTH WAS CROWDED AND VERY HOT.

In the exchange with HiSilicon related personnel, AI Nugget Found that the demand for back-end chips by domestic enterprises is still strong, but due to sanctions, HiSilicon can only consume stock, but it is obviously far from enough for the huge market, which is also one of the reasons for this year's chip fever.

In the article "Missing Core" upgrade of security, the eight reasons behind it, AI Nugget Zhi sorted out the reasons for the lack of cores from many aspects, and pointed out: Under the continuous lack of core crisis, the supply chain security problem is prominent, and the chip supply chain has become one of the core competitive elements of many enterprises in the security industry chain.

AI Nugget Believes that the essence of the lack of core security is the contradiction between the incremental demand and supply imbalance of high-end chips brought about by the upgrade of traditional security to AI security, and the US sanctions have exacerbated the outbreak of this contradiction.

To solve this "lack of core war", it is far from enough to rely on HiSilicon, and it is inevitable to the joint efforts of the entire industrial chain enterprises, on the one hand, to achieve domestic substitution, on the other hand, to promote the upgrading of the entire security chip industry.

From this point of view, the future "core pattern", whether it is front-end ISP/IPC, or back-end DVR/NVR, chip localization should be the mainstream trend. AI Nuggets also hope that in the future, domestic companies can take advantage of the momentum to support the security chip.

5

Keyword five: scene digitization

After the retreat of 360 Vision, the main force was placed on the digital city plate.

Coincidentally, Ping An Technology's original intention in transforming into a consulting company also saw the opportunities brought about by industrial digitalization.

For security companies, digitalization has worn everyone's ears out of the cocoon, in December 2021 AI Nuggets hosted the artificial intelligence security summit, Hikvision Li Yaya, Uniview Zhang Pengguo, Seagate Sun Dan and other leading corporate executives all talked about digitalization.

As Sun Dan said: The epidemic has pressed the pause button on some industries, but digital transformation is accelerating.

How to understand digitalization?

Zheng Zhibin, vice chairman of the China Cyberspace Security Association and former president of Huawei's Global Smart City Business Unit, said: Digital transformation is not a simple upgrade of informatization, but a new model for solving problems in the physical world.

In the context of digital transformation, "cross-field data fusion application scenario X goes to the deep water area".

Therefore, AI Nugget Zhi uses "scene digitization" to summarize the deep integration between technological transformation and actual landing scenarios.

Whether it is Haidaewoo or a number of AI companies, while mentioning industrial digitalization, it is inevitable to emphasize the landing scenario.

On the one hand, industrial digitalization covers a very wide range of industries, the business scope is very wide, different pain points between different industries, a single technology can not solve the problem, but requires a variety of technologies and even customized development to complete.

On the other hand, AI technology from the castle in the air to the landing, how to do a good job of combining technology and scenarios, is the top priority of AI companies.

In short, in the face of the historical opportunity of digitalization, all enterprises are doing scenario adaptation, covering as many scenes as possible with technology to establish their own moat.

AI Nuggets believes that the future competition will be based on scene-based competition. Whoever can deeply understand the scene and propose scenario-based solutions will be able to gain a firm foothold in the wave of digitalization and survive the fierce competition.

6

Keyword six: urban AIoT

Under the wave of digital transformation, the "de-security" of the security circle has become a consensus, and enterprises are now more willing to mention the concept of "urban AIoT".

Urban AIoT can be simply understood as a combination of "AI + IoT" (technology + scenarios) to solve various problems in the construction of digital cities.

Fundamentally, AIoT is the technology path for the digital transformation of cities.

Megvii senior vice president Cao Zhimin told AI Nuggets that AIoT is a technical means, and AI technology will distribute the data collected by sensing devices in various IoT scenarios to achieve digitization by machine means to solve the scene pain points, which is the value point of technology.

The reason why urban AIoT and digitalization are separated, AI Nuggets believes that digitalization is a major trend, while urban AIoT focuses more on urban governance, customer attributes are biased towards the G-end, and digitalization itself also includes B- and C-end services.

On the battlefield of urban AIoT, two routes have been formed: vertical and horizontal.

Hezhong emphasizes specialization, establishes ecology, and does deep work in some areas; lianheng focuses on diversification and pays attention to expansion, aiming to become bigger and stronger.

But in the view of AI nuggets, no matter whether the black cat is a white cat, it is a good cat to catch a mouse. For the combination of vertical and horizontal, different enterprises choose to conform to their own development path, are reasonable, the key lies in technology and products, whether they can really solve the problem, so as to form a positive cash flow closed loop, which is the basis for the sustainable development of enterprises.

8

Summary & Outlook

Looking back at 2021, the AI security circle seems to have changed, or the same wave of people, doing the same thing.

The difference is that under the change of the general environment, the free flow of enterprises, capital, talents and other elements.

If you collect these important elements and do a data analysis, you may be able to conclude that all kinds of elements are adjusted within the local area, but the whole circle is expanding.

In other words, the elements of the original circle are just rearranging and combining, but this brings new chemistry, making the business broader and the market wider.

In 2022, the enterprises and talents in the circle will enter a new journey.

END