Nvidia

Nvidia is one of the leading manufacturers of application smartphones and tablet processors, as well as GeForce mobile processors.

The company is the inventor of GPU, a high-performance processor that generates immersive interactive graphics on workstations, personal computers, game consoles and mobile devices.

NVIDIA's main office is located in Santa Clara, California, with branches in Asia, Europe and America.

Products and Services

Nvidia delivers its GeForce GPU products to the entertainment and consumer markets, Quadro GPU products are for the professional design and visualization market, and Tesla computer solutions are delivered to the high performance computing market.

Performance indicators

Main article: Nvidia financials

History

2024

A scrap of new processors that leads to overheating and can cause a fire in the data center

In mid-November 2024, it became known that Nvidia's Blackwell generation AI accelerators, when used in high-density server racks, could experience overheating. This creates a fire risk in the data center. Read more here

Intel leaves the Dow Jones index. Nvidia will take its place

On November 1, 2024, it became known that Intel's place in the prestigious Dow Jones Industrial Average index would be taken by Nvidia. This event underscores the shift in the chip manufacturing market in favor of AI solutions and is another blow for Intel, which has been part of the named index since 1999. Read more here.

Closing Russians access to new drivers for video cards

On October 23, 2024, it became known that the American company Nvidia deprived users from Russia of the ability to download new versions of drivers for their graphics accelerators. You cannot update drivers already installed on your computer.

Reports of problems with downloading Nvidia drivers come from various parts of Russia. Some users complain that they cannot go to the Nvidia website. Others say they have a driver download page open, but there are various errors when trying to get the file. In particular, a notification is displayed that access is denied - the request is blocked by the Edgecast WAF firewall. Users point to the following errors:

𝓲

Unsplash

• Access Forbidden. This request has been blocked by Edgecast WAF - access denied. Request blocked by Edgecast WAF;
• Access denied, in accord with Nvidia Information Security Policy - Access denied in accordance with Nvidia's information security policy.

It is also not possible to download drivers through the GeForce Experience application. Some owners of Nvidia video cards from Belarus also note that it is not possible to download drivers from the manufacturer's official website. Users complain that they have to download drivers for graphics accelerators from third-party resources. But in this case, there are certain risks associated with cybersecurity. However, in other countries such problems are not observed.

What exactly the restrictions are connected with is not clear as of October 23, 2024. Nvidia itself does not comment on the situation in any way. It is said that in the summer of 2022, the US authorities informed Nvidia about the introduction of new export requirements for the supply of the company's products to China and Russia.^[1]

China bans local companies from purchasing any Nvidia AI processors

At the end of September 2024, it became known that the Chinese authorities are increasing pressure on local companies, requiring them to purchase locally produced AI processors instead of Nvidia products. This decision is part of a set of measures by China to expand its semiconductor industry and counter US sanctions.

Therefore, Chinese regulators are demanding that local companies refrain from purchasing Nvidia H20 accelerators, which are used to develop and launch AI models. The move is supposed to help Chinese chipmakers increase market share while preparing local tech companies for possible new U.S. sanctions. Leading AI processor manufacturers in the country include Cambricon Technologies Corp. and Huawei Technologies Co. Earlier, Beijing also recommended that local electric car manufacturers purchase more parts from local manufacturers so that China can fully supply itself with all critical technologies.

𝓲

Nvidia

The U.S. government previously banned Nvidia from selling the most advanced processors to Chinese customers to limit Beijing's technological advancements. Nvidia modified subsequent versions of the chips so that they could be sold in accordance with US Department of Commerce rules. The H20 range meets these criteria. At the same time, Nvidia chips are the gold standard for companies that develop AI services, notes. Bloomberg According to the news agency, several Chinese technology companies, including ByteDance and, stocked up Tencent Holdings on Nvidia chips before export control measures came into force.^[2]

Nvidia employees work seven days a week until the night. They are promised millions for this

At the end of August 2024, it became known that many Nvidia employees work seven days a week, often until 1 or 2 am. But, despite the high workload, workers are in no hurry to leave the company, since they are actually in "golden handcuffs" - they are promised millions.

In particular, one of Nvidia's marketing employees said that she often had to attend from 7 to 10 meetings a day, each of which involved more than 30 people. And a former Nvidia corporate customer support worker said some of the company's meetings were accompanied by screams and even scuffles.

𝓲

Bloomberg

Nvidia has a relatively low level of personnel turnover. The incentive for staff to stay in the company is grants in the form of shares. From 2019 to 2024, the value of Nvidia securities rose by 3776%, which means that many employees who own the company's shares have become millionaires. In 2023, about 5.3% of employees left Nvidia. But after the company's market capitalization exceeded $1 trillion, the pace of layoffs fell to 2.7%. In the semiconductor industry as a whole, staff turnover is at 17.7%, according to Nvidia itself.

Those who have worked for the company for about a decade have more than enough money to retire, according to a former Nvidia engineer who left the company in June 2024. However, many do not do this, as they expect to receive another grant. Nvidia's success is primarily due to the rapid development of artificial intelligence: the company is a leading supplier of GPU-based accelerators for resource-intensive computing loads associated with AI. Demand for such accelerators is driving growth in Nvidia shares.^[3]

2023

Canceling orders from China

On October 31, 2023, Nvidia shares fell 5%. This is due to the fact that the company canceled the order of chips for $5 billion.

Orders that Nvidia planned for 2024, namely the supply of chips to giants such as Alibaba, ByteDance (owner of TikTok), Baidu, fell under the sanctions of the US Department of Commerce.

Nvidia shares fall 5%

Nvidia's share price fell to $392.3, its lowest since July 2023. If you count from a record high price - $493.55 apiece in August 2023, then the value of the shares fell by 20%.

Nvidia said that chips are always in high demand, and it takes a lot of time to produce them. Nvidia is working hard to send numerous orders to its customers in the U.S. and beyond, the company said.

The new export rules will not have a serious impact on this quarter's performance, Nvidia's press service said.

Earlier, Joe Biden the administration introduced new export bans related to new chips. Such measures are taken in order to cut China off from American advanced technologies that can strengthen the Chinese army. Mainly, the latest generations of processors are used to create models. artificial intelligence Last year, the United States already banned Nvidia from selling H100 chips to China, but the list of models banned for export has expanded. Most likely, according to sources, the Bloomberg ban extended to fast H800 and A800 processors.

The rules came into force in November 2023. In addition to China, Iran and Russia also fell under the export ban.

China condemns U.S. restrictions on semiconductor exports. On October 16, a spokeswoman for the Chinese Foreign Ministry, Mao Ning, said that Beijing opposes "politicization, turning into an instrument and into a weapon of trade and technology" by the United States].^[4]

Searches of offices in the case of conspiracy in the chip market

At the end of September 2023, it became known that the French antimonopoly regulator conducted searches in local offices. Nvidia The world's largest supplier of accelerators for systems artificial intelligence is suspected of anti-competitive behavior.

The raid, authorized by the court, was reportedly of a sudden nature. According to The Wall Street Journal, law enforcement officers arrived at Nvidia's French office early in the morning to quickly interview arriving employees. Such raids provide for the possibility of withdrawing physical materials and digital documentation. French authorities say they conducted the searches "as part of an increased focus on cloud technology." Nvidia has no comment on the situation as of early October 2023.

Nvidia offices searched

Nvidia dominates the market for high-performance graphics accelerators (GPUs) used in machine learning applications such as neural networks and large language models. Against this background, French regulators suspect Nvidia of violating antitrust laws when selling GPU-based products.

It is also noted that the raid is part of a broader investigation into cloud computing, which the French authorities began in January 2022. According to The Wall Street Journal, regulators have targeted Nvidia for the reason that demand for the company's AI accelerators is growing rapidly. Nvidia itself says Alibaba, Amazon, Google, Meta services (recognized as an extremist organization; activities in the Russian Federation are prohibited), Microsoft, Snap, Spotify, Tencent and 40 thousand other companies are created and operate on Nvidia AI technologies. No official charges have been filed against Nvidia as of early October 2023.^[5]

2022

Closing office in Russia

On October 3, 2022, it became known about the imminent closure of Nvidia's representative office in Russia. About 240 people remained in the Russian office of the company, who are expected to move to other countries.

Nvidia has officially left Russia, follows from its reporting for the third quarter of 2022.

The fact that the Russian office of Nvidia will stop working by the end of October 2022 was reported by Forbes magazine, citing a source and a company representative. According to the informant, the manufacturer of graphic solutions by the beginning of October 2022 "actively takes out on charter aircraft those who agree to be released to offices in other countries."

Nvidia closes office in Russia

An Nvidia spokesman did not comment on the information about the export by charter flights, but, according to him, Nvidia announced to employees on September 30, 2022 that it was closing its office in Russia due to "the inability to ensure their effective operation." Employees who do not want to leave Russia will be terminated by agreement of the parties with severance pay, Nvidia said. What will be the amount of compensation is not specified by October 3, 2022.

Until February 2022, more than 300 employees worked at Nvidia in Russia, and a few months later their number decreased by about 60 people, mainly engineers, a company representative told the publication.

In April 2022, Nvidia announced that it would create a research center in Yerevan. In early September 2022, the company announced that the United States had banned it from exporting two types of processors designed to speed up machine learning tasks to Russia and China. The US fears these processors could be used or redirected to "military end-use" or "military end-user" in China and Russia.^[6]

US officials ban the company from exporting two of the best chips to Russia and China

In August 2022, Nvidia Corp reported that US officials told the company to stop exporting two of the best AI computing chips to Russia and China, which could undermine the ability of Chinese firms to perform advanced work such as image recognition and complicate Nvidia's business in China.

Nvidia said U.S. officials told it the new rule "would eliminate the risk that the products in question could be used or redirected to" ultimate military use "or" ultimate military user "in China."

According to the new license requirements, restrictions apply to GPU-based computing accelerators (graphics processing unit) that meet certain performance criteria, such as A100 and H100 series products. Such models are used to work with artificial intelligence and increase computing performance at times.

Stop selling cloud software licenses in Russia

Nvidia has stopped renewing and selling licenses to Russian companies for its cloud gaming software. The decision was made against the background of the ongoing conflict in Ukraine.

Nvidia decided not to close the development center in Russia

As it became known to TAdviser on June 16, 2022, the American technology company Nvidia does not plan to close the Russian development center. TAdviser was informed about this by the company.

Nvidia's business is out of politics as far as it can now, and we have never announced a departure. At the same time, deliveries of products that are produced directly under the Nvidia brand have been stopped: Founders Edition video cards, Nvidia Shield TV consoles, as well as Nvidia solutions for professional and corporate markets. It is worth noting that the Moscow office of the company did not open after closing during the pandemic. All this time, office employees work remotely, - said the Russian press service of Nvidia.

Nvidia does not plan to close the Russian office, but stopped selling products in the Russian Federation. The photo is lookatme.ru

Manufacturing companies make their own decisions on the supply of products based on Nvidia technologies for both the mass and corporate markets. That is, manufacturers - such as HP, Dell, Lenovo, Asus, Gigabyte, MSI, etc. - themselves decide to supply or not supply laptops, computer and server equipment or mining equipment created using GPUs, video cards and Nvidia technologies to the country. Moreover, such global services as AWS from Amazon and Unreal Engine from Epic Games, based on Nvidia technologies, also decide on availability in Russia, the developer does not impose any obligations on them.

One of Nvidia's Russian partners clarified that there are no problems with the hardware from Nvidia itself, "alternative supply channels" have been launched, so video cards and GPUs are available to domestic customers.

In April 2022, it became known^[7]Nvidia was opening a research center in Yerevan (Armenia). As the company explained to TAdviser, the Armenian center was created not to replace the Russian one, but in addition.

Those employees of the company who expressed a desire to move to Yerevan moved, the rest continued to work in the Russian office, the press service said.

At the same time, as it became known to TAdviser from one of the company's employees, the Russian office of Nvidia continues to develop. The staff of the company has been formed, and there is no mass hiring in Russia, and if negotiations are underway with someone, then these are isolated cases, Nvidia said.

At the same time, there are no ads for hiring employees on the Nvidia global website or in Russian recruiting services. The global website contains more than 2.8 thousand vacancies around the world, most of all vacancies are open in the United States - almost 1.5 thousand, India - 455, Israel - 299 and China - 269. 80 vacancies have been officially opened in Armenia.

In 2003, Nvidia opened a laboratory in Russia to test computer games. The press service of Nvidia refused to answer questions about what the Russian office is now specializing in, what the Armenian office will do, how many employees work in Moscow and how many are planned to be hired in Yerevan.

A source in Nvidia shared with TAdviser that several hundred people work in the Russian office who are engaged in software development, machine learning (ML), technology consulting, including testing Nvidia technologies in various development environments.

Stopping supplies to Russia

American chip developer Nvidia will suspend sales in Russia, a company spokesman said, PCMag wrote on March 6, 2022. According to the newspaper, this happened due to the Russian military operation in Ukraine.

The company clarified that the suspension of sales will affect all Nvidia products. "We are not selling to Russia," the spokesman said.

Hacker attack

At the end of February 2022, the internal network Nvidia was hacked, and several key systems, such as email and developer tools, were down for several days. According to CRN, a South American group hackers called LAPSU $ hacked into the company's internal network and copied more than 1 TB of critical company data.

Nvidia continues to investigate the incident as of February 28, 2022, and its response to the CRN request says that business and commercial activities continue continuously. The company is still working on assessing the nature and scale of the event and has no further information. Presumably, the hackers were able to get inside the perimeter of Nvidia after gaining access to the email accounts of some of the company's employees.

Nvidia attacked by hackers

According to insiders, after hacking, penetrating the perimeter and downloading more than 1 TB of data, LAPSU $ itself was subjected to a reverse attack by Nvidia information security (IS) specialists. Information security specialists of the company managed to penetrate the computer systems of the LAPSU $ hacker group and encrypt all stolen data and other information on hackers' personal computers (PCs). However, the hackers still have a leak backup on the virtual machine, where Nvidia specialists could not penetrate.

LAPSU $ began distributing the obtained information in a closed community, for example, hackers published the source code of Nvidia drivers. Experts explained that until 2022, IT companies did not act so aggressively as part of protecting their data. Usually, business companies either came to terms with the loss of data, or paid hackers a ransom so that attackers would not publish stolen information^[8]

Nvidia ARM deal fell through due to regulators

On February 8, 2022, the deal to sell Arm to Nvidia was officially announced. The latter will receive compensation from SoftBank (owner of Arm) in the amount of $1.25 billion for the fact that the takeover of the British chip developer did not take place. Read more here.

Purchase a software developer to manage hundreds of Bright Computing servers

In early January 2022, video card maker Nvidia announced the purchase of Bright Computing. The financial component of the transaction of the company did not disclose. Read more here.

2021

Buying a developer of maps for unmanned cars DeepMap

In mid-June 2021, the chipmaker Nvidia announced the purchase of a card developer for. unmanned vehicles DeepMap Financial terms of the transaction are not disclosed. The deal is expected to close in the third quarter of 2021. More. here

From GPU development to a comprehensive AI infrastructure for data centers. Main announcements of GTC 2021

Opening the annual NVIDIA GTC conference on April 12, which was held online for the second year in a row due to the pandemic, NVIDIA CEO Jensen Huang made a number of important announcements. The company is actively increasing the presence of its technologies in all industries and at all levels - from PCs to data centers and clouds. NVIDIA also continues to democratize artificial intelligence ( AI) technologies, making them easier to access and cheaper to use for both individual researchers and large companies. Read more here.

2020

ARM Purchase

On September 14, 2020, Nvidia announced the acquisition of ARM from Japanese holding SoftBank for $40 billion. Under the terms of the deal, which its participants want to close by the end of March 2022, Nvidia will pay $21.5 billion with its shares and $12 billion with its own funds, including $2 billion immediately after the agreement. Read more here.

ARM Acquisition Plans

On July 31, 2020, it became known that Nvidia is considering buying a British chip developer, ARM, owned by the Japanese corporation SoftBank Group. Read more here.

Purchase of OS developer for Cumulus Networks switches

In early May 2020, Nvidia announced the acquisition of Cumulus Networks. For how much the OS manufacturer was sold for data center switches, not reported. Read more here.

Plans to complete the acquisition of Mellanox

On April 24, 2020 TAdviser , it became known about the corporation's plans Nvidia to complete the transaction in the near future to acquire all the assets of the Israeli developer of network equipment and solutions for. data centers Mellanox Announced back in March 2020, the deal recently received the approval of the authorities. Its China closure should take place on April 27, 2020. More. here

2019

Among the largest companies in the semiconductor industry by revenue

Крупнейшие компании в индустрии semiconductors by revenue in 2019

Purchase of chip developer for supercomputers Mellanox $6.8 billion

On March 11, 2019, Nvidia announced the acquisition of supercomputer chip developer Mellanox Technologies for $6.8 billion. With this deal, the buyer will be able to expand his rapidly growing data center solutions business. Read more here.

2017

Refusal to support 32-bit OS since the beginning of 2019

In December 2017, Nvidia posted a statement on the technical support page of its website to discontinue the release of drivers for its graphics cards for all types of 32-bit operating systems.

Thus, after the release of the next universal driver package Release 390, the company will stop releasing updates for 32-bit systems for any type of graphics accelerators

Later versions of the drivers will not work or be installed on 32-bit operating systems. Optimization, extensions and other driver options after Release 390 will not support 32-bit compatibility with Release 390 or earlier, the company said in an official release.

All options are included in the list of 32-bit operating systems that will lose further support for Nvidia Linux and FreeBSD, as well as MicrosoftWindows 7, Windows 8/8.1 and Windows 10.

According to the ArsTechnica portal, the official release of the next universal Nvidia driver package Release 390 is expected in January 2019. Until then, Nvidia intends to continue releasing critical security updates for outdated 32-bit operating systems.^[9]

Partnership with Baidu to introduce AI into clouds, self-driving cars and smart homes

Nvidia and Baidu partnered in early July 2017 to introduce artificial intelligence (AI) into cloud computing, self-driving cars and smart home systems.

Baidu President and Managing Director Ki Lu (Qi Lu) presented the following plans for cooperation with Nvidia:

• Deploying next-generation Nvidia Volta GPUs to Baidu Cloud to create a deep learning platform

• introducing the Nvidia Drive PX platform into Baidu's autonomous driving system and developing self-driving vehicles in conjunction with key Chinese automakers;

• optimization of the open framework of deep learning Baidu PaddlePaddle for GPU Nvidia Volta and its distribution in the scientific environment;

• providing AI capabilities to Chinese consumers by adding support for Baidu DuerOS-based OS in Nvidia Shield TV.

"The result of this collaboration will be the creation of AI-based computing platforms for both research and startups working on advanced AI applications and autonomous cars," said Ian Buck, vice president and director of accelerated computing at Nvidia.

"We are pleased to announce the beginning of a versatile partnership with Nvidia. Baidu and Nvidia will work together on the Baidu Apollo autonomous driving platform, integrating Nvidia's automotive technologies into it. We will also work together to make PaddlePaddle the optimal deep learning framework, optimize our DuerOS AI-based OS, and accelerate research at the Deep Learning Institute, "Ki Lu commented on the event

AI in the cloud and data centers

Baidu intends to deploy Nvidia HGX architectures with Tesla Volta V100 and Tesla P4 accelerators to train deep networks and inference in data centers. As conceived by the company, scientists and researchers armed with the Baidu PaddlePaddle deep learning framework and the Nvidia TensorRT inference program will be able to create products and services with real-time recognition of images, speech, text and video.

To develop AI technologies, companies will jointly optimize the open-source Baidu PaddlePaddle deep learning framework based on the Nvidia Volta architecture. The source code for the framework was opened in 2013. PaddlePaddle is used to create Baidu solutions, including Baidu search ranking, image classification, real-time speech recognition, visual pattern recognition, and other AI-based services from third-party developers.

AI in self-driving cars

Baidu intends to use the Nvidia Drive PX 2 supercomputer for its autonomous driving platform. Nvidia and Baidu are already collaborating on autonomous driving under the Baidu Apollo program.

Apollo, an open platform for self-driving cars, uses Nvidia's entire suite of technologies - from server to car, including Tesla's GPU and Nvidia Drive PX 2, along with Nvidia's software, including CUDA and TensorRT.

The newly introduced Baidu self-driving car demonstrated the capabilities of the Apollo system powered by the Drive PX 2.

Several Chinese automakers have already announced their entry into the Apollo alliance, including Changan, Chery Automobile, FAW and Greatwall Motor.

AI at home

According to Baidu's plans, DuerOS, an AI-based operating system, will bring voice command support to Nvidia Shield TV in the Chinese market. Shield is a solution for streaming, gaming and smart home. A version of the console with software adapted for China will enter the market before the end of this year, the company said.

Funding from the U.S. Department of Energy on HPC

Nvidia became one of the six companies that received funding from power engineering specialists USA the Ministry in the amount of 258 million dollars under the Project on Exaflops Computing (ECP - Exascale Computing Project) for the development of supercomputers a new generation.

See also: IBM, Intel, Cray will give out a quarter of a billion dollars from the treasury

The goal of the ECP project is to develop at least two exaflops computing systems, one of which should be ready by 2021. Such systems should be about 50 times more powerful than the most powerful supercomputer to date - the Titan system operating at Oakridge National Laboratory.

The ECP PathForward program is focused on finding solutions that ensure the most efficient use of energy at the highest possible performance for future supercomputers designed for applications in areas such as national security, manufacturing, and power research.

Developing supercomputers is a priority for the Department of Energy. In addition to a significant increase in productivity, the Department of Energy seeks to reduce energy costs as much as possible and achieve exaflop performance with energy consumption of 20-30 megawatts. To date, the CPU-based exaflops system consumes hundreds of megawatts of electricity.

For more than a decade, NVIDIA has been researching and building increasingly powerful and energy efficient GPUs for high-performance computing. This is the sixth subcontract with the Ministry of Energy. It will help accelerate the development of highly efficient computing technologies.

As part of the R&D initiative, the NVIDIA team will address critical tasks, including creating energy-efficient GPU architectures and developing fault tolerance systems. The results of the work will be used in future generations of GPU architectures that will replace Volta. In turn, solutions on Volta will form the basis of the flagship supercomputers of the US Department of Energy - Summit and Sierra - the launch of which is scheduled for 2018.

2011

Kal-El processor announcement

In the spring of 2011, Nvidia introduced a 4-core processor codenamed Kal-El, designed for the widest range of applications, including mobile devices.

Microsoft received the right to purchase Nvidia assets

In June 2011, it became known that Microsoft received the right to first acquire large stakes in Nvidia if they were sold. The existence of such an agreement, citing documents filed with the US Securities Commission, was reported by InformationWeek.

In the event that any person or any company makes an offer to Nvidia to purchase a number of shares equivalent to or exceeding 30%, Microsoft has the exclusive right to first repurchase this package, according to documents filed with the commission on May 27, 2011.

Thus, Nvidia cannot be sold in its entirety, except if Microsoft itself wishes to buy the company. This suggests that Microsoft may be interested in buying a promising chipmaker. According to Nvidia's current stock price, a third of its share capital is worth about $3.4 billion.

Microsoft's interest in the California company may be dictated by the software giant's desire to strengthen its position in the mobile segment. In early 2012, a release is expected that will operating system Windows 8 support ARM architecture for the first time, and Nvidia is just releasing such processors.

With its own chip developer and supplier, Microsoft will be able to offer comprehensive computing solutions, as the same Intel, which is actively involved in the development of open source operating systems such as MeeGo, strives to do.

If acquired by Nvidia, Microsoft will follow in the footsteps of Apple, which, after purchasing PA Semi in 2008, launched its own mobile processors - for the iPhone, iPad and iPod touch. Consolidation of diverse resources in one hand (including software development and component release) allows you to create more efficient products and increase their added value, experts say.

According to InformationWeek, apparently, an agreement on the purchase of Nvidia shares was reached by the parties more than 10 years ago as part of a broader agreement, which, among other things, determined the procedure for purchasing GPUs for the Xbox game console. In 2003, Microsoft switched to using ATI chips, but other conditions remained in force, the newspaper notes.

See also

• Supercomputers
• Supercomputers (global market)

Notes