The hottest NVIDIA plummeted. What happened to the

  • Detail

NVIDIA plummeted. What happened to the first stock of artificial intelligence

in the past month, as the most dazzling AI concept stock in the US stock market, NVIDIA (NVDA). The share price fell continuously, from the highest $292.76 on October 1, 2018 to the lowest $161.61 on November 16, 2018. The share price retreated by nearly 45%

as one of the most beloved companies on Wall Street and a necessary AI concept stock for institutions and retail investors, NVIDIA plummeted continuously after the latest quarterly report, creating a 52 week low. People can't help but wonder, what happened to NVIDIA? Is it possible that the artificial intelligence of the past and the tests we did later proved that it was cool

before answering what happened to NVIDIA, we need to popularize a concept to help us understand the era background of NVIDIA. That is, why NVIDIA, a GPU company, can become the first AI stock in the market in recent years

what is the business model of AI

the landmark rise of AI industry was undoubtedly the epic duel between alpha go of deepmind team and go world champion Li Shishi at that time. At that time, everyone was not optimistic about alpha go, but the result was unexpected. Alpha go lost 4-1, and Li Shishi won

when people marvel at the power of alpha go, smart businessmen also see the business opportunities that can change the world. Soon, following closely, a large number of companies engaged in artificial intelligence have sprung up. So how does AI work? For the convenience of understanding, we can simply split it into three steps:

1. We have obtained a large amount of raw data

2. We input these raw data into the artificial intelligence algorithm through the artificial intelligence algorithm engineer

3. With the support of a large amount of computing power, artificial intelligence algorithms process these data, output them into more intelligent information, and apply them to terminal products to make terminal products more intelligent

therefore, in these three steps, we can see that AI has three core elements that are essential:

1, a large amount of raw data

2, Algorithm Engineer

3, a large amount of computing power

and among these three core elements, NVIDIA happens to occupy the two core elements of algorithm engineer and computing power

in terms of computing power, the GPU produced by NVIDIA is more suitable for the large-scale general parallel computing power required by AI than CPU. So many AI companies buy NVIDIA GPU as AI training chip

on the other hand, NVIDIA invested a lot of time and money to cultivate CUDA compiler, which occupies a very high market share in the use scenarios of GPU engineers. Successfully overcome the technical difficulties of 3D printing PLA line materials, such as brittleness and temperature resistance. For these Algorithm Engineers, getting used to CUDA compilers has also become a moat at the software level of NVIDIA

therefore, to some extent, NVIDIA has a first mover advantage and monopoly effect in the field of artificial intelligence. Basically all AI related companies will purchase NVIDIA's products at present

naturally, NVIDIA has also transformed from a hardware manufacturer making game chips to a technology company now with AI as its core competitiveness. In its financial report, the business is also divided into five areas: Games, professional visualization, data center, autonomous driving, licensing and OEM

in the eyes of the market, from the moment of NVIDIA's successful transformation, NVIDIA has been given the concept of the first stock of artificial intelligence and has become the darling of Wall Street. Under the double-click of Davis' profit and performance growth, NVIDIA's market value has risen steadily, becoming a very typical crowd trading

there is a saying that there is only a gap between genius and madness. In fact, in the capital market, from Davis' double-click to Davis' double kill, it is only a blink of an eye

what are the characteristics of crowded trading? One is that investors' risk appetite has gradually increased, and the valuation multiples of enterprises have continuously increased. This can be seen from the historical changes of NVIDIA PE

both success and failure. The high valuation has become the root cause of NVIDIA's recent stock price decline. So from the result, what is the catalyst that caused this event? We can find the reason from the latest financial report released by NVIDIA

start with NVIDIA's latest quarterly report

NVIDIA released its latest quarterly financial report after the US stock market closed on November 15, 2018. The report showed that the total revenue was $3.181 billion, a year-on-year increase of + 21%, but the month on month increase was negative. Moreover, the expected performance outlook for the next quarter is $2.7 billion, with a month on month negative growth of nearly 17%

investors have given NVIDIA a very high growth expectation and valuation level. Such poor performance expectations undoubtedly gave investors a heavy blow

in NVIDIA's financial report, the management attributed the slowdown in revenue to the backlog of GPU channel inventory of mid range Pascal products. The underlying cause of channel inventory backlog is the decline of digital currency

mining bitcoin requires a miner, and the miner needs computing power to run the hash algorithm, which comes from the purchased NVIDIA GPU. The hot digital currency market has given NVIDIA a huge demand market for GPU retail business, but once the digital currency market falls, a large number of mining machines will be eliminated if the mining cost is higher than the digital currency price. These eliminated GPUs entered the second-hand market, which not only led to the continuous decline of GPU prices, but also completely saturated the short-term market demand, which had a negative impact on the production capacity of upstream NVIDIA

data source: (price unit: USD)

BTC price trend

so how long will this impact last? According to the management of NVIDIA, the expected inventory backlog may take 1-2 quarters to be released. However, from a more conservative point of view, if the digital currency falls further and operates at a low price for a long time, more second-hand GPU increments will be released, which is undoubtedly a fierce battle about time and price for NVIDIA

in the same situation, we can also learn from the relationship between the oil service industry and crude oil. Crude oil prices have continued to fall since 2008. Although they bottomed out in 2016, the initial excess oil service capacity has not improved due to the recovery of oil prices. Taking ETF of oil service industry as an example, it can also be seen that the fluctuation and rebound of oil price do not change the continuous decline of oil service

what about NVIDIA GPU? From the perspective of performance split, NVIDIA's retail GPU sales revenue is mainly divided into the field of game revenue. The whole hot digital goods trading price in 2017 has not been announced, which has driven the booming development of the mining market. It can also be seen from the quarterly revenue growth that the game business in the first quarter of fiscal 2019 increased by nearly 68% year-on-year compared with the first quarter of fiscal 2018

the game business is still the main source of revenue for NVIDIA's entire revenue segment, accounting for about 50% of revenue. The rapid growth of revenue has supported NVIDIA's high valuation. However, once the digital money market bears, prices fall, and a large number of second-hand GPUs continue to flow into the market, leading to the shrinking market demand of NVIDIA's game business, slowing down the growth of the game business, which accounts for 50% of the revenue, and even further declining to the $1 billion revenue level of q1fy18, it will undoubtedly be a fatal blow to NVIDIA's valuation system, which is also an important reason for the poor performance outlook of NVIDIA

outside the game

in addition to the game business, the two businesses that investors pay more attention to are autonomous and date center

in NVIDIA's valuation system, data center and autonomous driving have always been important businesses supporting NVIDIA's belief in artificial intelligence. Because they are the most landing business scenarios in the field of artificial intelligence. However, according to the latest quarterly financial report of the Department of raw materials industry, which firmly focuses on the work of the Party group center of the Ministry, the growth data of the data center is slowing down, and the time point for the large-scale growth of autonomous driving is still in the future

data center is currently the second largest revenue source of NVIDIA after games. How to understand this business? It can be simply understood that other technology companies are making artificial intelligence cloud computing platforms and need to purchase NVIDIA's GPU. For large companies such as Google, Amazon and IBM, NVIDIA's revenue in the data center corresponds to the cost of capital expenditure of these companies

the month on month growth rate of the data center has increased from 16% in the previous quarter to 8% in the previous quarter, and then gradually declined to the latest 4% month on month growth. Perhaps two problems can be seen:

1. The phased growth of AI market capacity has slowed down. It is possible that these large companies' investment in data center servers is slowing down prudently before the large-scale commercial landing of AI

2. Competitors in the same industry began to enter. In particular, Huawei is also launching its own AI chips to negotiate and cooperate with corporate customers such as Microsoft. Google's TPU is also entering the market of artificial intelligence chip training end. Logically speaking, cloud computing has the competitive characteristics of economies of scale, and being able to reduce its own costs is the demand of these companies that need to establish data centers. Moreover, when artificial intelligence erupts on a large scale, these technology companies will certainly not be helplessly watching NVIDIA enjoy such a huge market. This also makes investors worry about whether NVIDIA can still enjoy this high-profit cake alone

in NVIDIA's self driving business, the main problem is that the current volume is small, with a single quarter revenue of only $172 million, accounting for 5.4%. From the perspective of large-scale growth, the market is expected to be 2020. For example, Volvo, the traditional car factory that cooperates most closely with NVIDIA, will also be able to mass produce L2 autonomous vehicles based on NVIDIA Xavier scheme in 2020. Or as the market believes, 5g communication can bring fundamental changes to automatic driving. In short, the current self driving business is not enough to become NVIDIA's potential performance growth point, and there is still great uncertainty between 2020 and 2020

so to sum up, NVIDIA's current situation - affected by the concept of artificial intelligence, NVIDIA's short-term overvaluation has become the fundamental reason for the stock price adjustment. In terms of performance, investors began to worry about NVIDIA's future performance growth capacity because of overcapacity caused by the decline of digital currency, as well as the slow growth of data centers and the slow progress of autonomous driving

value or price

back to the source of investment, the relationship between price and value. Graham said that the stock market is a voting machine in the short term and a weighing machine in the long term. Most of the time, the stock price fluctuates around the stock value

from the perspective of current price, NVIDIA's static PE valuation is 22 times, which seems relatively fast corresponding to the year-on-year growth rate of 20% in the latest quarter

Copyright © 2011 JIN SHI