Posts

Cloud vendors go deep with industry solutions

Google Cloud and SAP are flagship examples of vendors building vertically oriented strategies with the introduction of industry cloud solutions

Over the last year, as more organizations accelerated the migration to the cloud, it has become evident that a horizontal cloud does not always meet the specific needs of certain industries, especially highly regulated ones such as banking or healthcare. In response to the increasing complexity of certain industries and the regulations that they are governed by, along with the overall uptick in cloud adoption during the era of COVID-19, cloud vendors are investing in the development of clouds that cater to industry-specific nuances and the demands of safely distanced or remote work. The narrowing of focus is part of broader strategic objectives that extend beyond just the immediate commercialization of opportunity and revenue impact.

Initiatives by SAP and Google highlight how innovation and technological advancements in AI and machine learning (ML) factor into the evolution of industry clouds. With recent rollouts and announcements, both vendors have highlighted the importance of creating new types of data-driven cloud solutions powered by AI and ML, augmented by networks of customers and partners. While SAP and Google currently participate in fruitful partnerships together, it will be interesting to watch the common goal of delivering vertical-specific capabilities to their customers unfold to see whether the two companies align or compete in this new chapter of cloud defined by industry.

Google leads with innovation in ML and AI to augment, rather than replace, legacy solutions

In February Google Cloud CEO Thomas Kurian emphasized that Google Cloud’s top priority is to develop a new generation of applications, which will be able to extract data from traditional line-of-business (LOB) applications such as those used for CRM and use technologies like AI and ML to optimize outcomes. Kurian reinforced the strategy of veering away from the development of traditional enterprise applications in favor of industry-specific apps that are designed to extract data from these traditional applications.

Leading up to 2020, the migration timelines for the 80% of companies that had not yet made the move to the cloud rested on comfortably planned milestones that indicated an evolution versus an urgent call to change. Looking back over the last six months, organizations suddenly found themselves in the thick of the impetus to migrate to the cloud and to do so quickly. Even organizations whose cloud journeys were well underway prior to the COVID-19 outbreak are narrowing their strategies and turning to clouds that cater to the specific requirements and compliance standards that industries, especially those that are highly regulated, including the government, require. As a result, a number of leading cloud vendors such as IBM (NYSE: IBM), Google (Nasdaq: GOOGL) and SAP (NYSE: SAP) have recently made strides in building out and refining their cloud strategies to cater to the specific requirements and demands of certain industries.  

East meets West: A comparative tale of two e-commerce giants placing big bets on the cloud

Alibaba Cloud, AWS focus on build-outs of global footprints via infrastructure investments in 5G and expansion of data center and edge locations

While the growth of the two globally dispersed e-commerce giants Alibaba and Amazon is largely fueled by retail, both businesses have showed marked dedication to the growth of their respective cloud empires, focusing on infrastructure to fuel global expansion and investment in augmenting their respective portfolios. The investment in cloud is evident as the backbone driving each business as they compete on the global stage to become leaders in digital transformation (DT).

Alibaba Group’s aforementioned profit margins, fueled by its B2B operating model, have enabled a hefty investment of $28 billion dedicated to the cloud business. As the world was gripped by the initial effects of the COVID-19 pandemic back in April, the parent company announced the allocation of the sum, which stands as a massive proclamation of the company’s dedication to cloud and related technologies as the core drivers toward the enablement of DT. The sweeping investment, coupled with new leadership and an expanding partnership strategy, solidifies Alibaba’s intent to position its cloud business as a viable contender against AWS, especially in APAC. Alibaba is clearly placing a large majority of its bets on cloud and the future of DT, as the investment equates to 40% of its total 2019 revenue and is 5.7x the revenue of Alibaba Cloud.

The investment will have a profound impact on Alibaba Cloud’s ability to execute on strategies around DT, infrastructure build-out and R&D, and came at a time when the world could not have been in more need of capabilities such as increased bandwidth and enterprise and SME support. The focus in the medium term is a multifaceted push to gain scale globally, attract new customers and expand wallet share with existing customers, and much of this growth will be propelled by the expansion of Alibaba Cloud’s infrastructure backbone with the build-out of data centers and investment in 5G across EMEA and APAC.

Since solidifying its dominance in China and garnering competitive positioning on the global stage, Alibaba has been frequently referred to as the “Amazon of China.” Both companies have anchored their businesses as e-commerce platforms and have demonstrated parallel growth trajectories, becoming mainstays in the lives and businesses of customers globally. The uniqueness of their respective journeys, which have been significantly shaped by their foundations as e-commerce giants, does not overshadow the companies’ similar strategies. Over time, Alibaba and Amazon have evolved rapidly into diversified companies with a distinct focus on technology and digital transformation. While the companies are in different phases of their growth, in terms of size and global footprint, and have different operating models, the investments in and focus on their respective cloud businesses to drive growth are evident when comparing their evolutions and forward-looking growth strategies.  

Zoom videoconferencing booms amid COVID-19, but Microsoft and Google will win with broader SaaS

Zoom’s rapid adoption highlights security concerns; Microsoft and Google are better positioned

Microsoft (Nasdaq: MSFT) and Google (Nasdaq: GOOGL) were relatively well prepared for the unexpected demand for videoconferencing, with global data centers and strong security measures in place, while Zoom’s security was not ready for the same scaling. There have been numerous reports of “Zoombombing,” which is when hackers join a Zoom meeting and disrupt the workflow, particularly in online classrooms. While some educators did not use password protection to make their online classrooms private, TBR believes this security lapse also occurred because Zoom lacks end-to-end encryption. Microsoft and Google Cloud have experienced difficulty with outages but have performed well in terms of security with end-to-end encryption in Teams and Meet. The regularity of these hacks has led to a recent investigation by the FBI and caused some government agencies, companies and educational institutions to ban the use of Zoom. In TBR’s special report Security measures taken to combat impacts of COVID-19 on businesses will have long-term implications, Senior Analyst Nicole Catchpole discusses the security concerns with Zoom and other cybersecurity threats that have risen amid the COVID-19 pandemic.

Zoom, Microsoft and Google remove pay barriers, increasing usage and setting the foundation for a much larger base of paying customers post-COVID-19

Use of videoconferencing solutions is skyrocketing, but modernization of these platforms will be a long-term strategy

Zoom, Microsoft and Google are offering their video-collaboration tools for free to support the unexpected global shift to remote work and learning environments. For six months, Microsoft is removing paywalls for Government Cloud and certain Office 365 subscriptions — including Office 365 E1 for businesses and Office 365 A1 for educational institutions, both of which include Microsoft Teams. Google Cloud is also offering Meet for free until Sept. 30, but only to existing customers, which makes it slightly more restrictive than Microsoft’s offer. Finally, Zoom has also removed the 40-minute time limit on its free basic subscription tier for K-12 schools in numerous countries, including the U.S., where the company boasts roughly 60,000 customers as of mid-March. Within the “freemium” tiers that are available, Zoom customers can have up to 100 participants in a virtual meeting, whereas Teams and Meet can support up to 250 people in a meeting. Given that each of these vendors has reduced cost-related barriers to adoption, customers can select the vendor that best fits their broader IT environment.

The global shift to a virtual work-from-home and learn-from-home environment has drastically increased demand for SaaS solutions that support collaboration and remote workflows, particularly videoconferencing as companies and educational institutions try to maintain as much face-to-face communication as possible. Among the numerous SaaS offerings available, some of the most popular are Zoom, Microsoft Teams and Google Meet. Zoom quickly rose in popularity and became a household name, growing from 10 million users in December 2019 to 200 million users in March 2020. While Zoom’s (Nasdaq: ZM) number includes individual nonpaying consumers, the vendor has also signed paying business and organizational customers including IAC Group, Rubrik and Texas A&M. Microsoft Teams and Google Meet experienced growth spikes as well. The number of users on Teams more than doubled, from 20 million daily active users in November 2019 to 44 million in March 2020, including enterprises such as EY, SAP (NYSE: SAP), Continental AG and Accenture (NYSE: ACN). Google Meet grew by more than 25 times from January to the end of March and is adding 2 million new users per day, with customers such as Korean Air, Shopify (NYSE: SHOP) and TELUS (NYSE: TU). While Zoom’s total user growth is strong and its offerings are widely used, TBR expects Microsoft and Google Cloud will start to poach Zoom customers due to their value-add hardware and SaaS offerings Office 365 (200-plus million users) and G Suite (6-plus million users).

AI, Accenture and Amazon: HITS acquisitions update 2020

Accenture’s steady appetite, Amazon’s potential new offering and Google’s uncertain moves

Accenture’s acquisition of Clarity Insights follows the company’s INTIENT purchase and rounds out a typically active acquisition year for one of the leaders in TBR’s HITS benchmark. Clarity Insights brings Accenture AI and machine learning capabilities, 350 healthcare data scientists, and healthcare industry clients. As noted in our most recent full report on Accenture’s HITS business, “Accenture targeted the AI opportunity in life sciences in mid-2019, launching its INTIENT platform for collecting, storing, monitoring and analyzing data from life sciences clients’ business environments. The platform leverages Accenture Applied Intelligence to provide AI and analytics services, improving efficiency and data management.” Beyond extending Accenture’s capabilities, the Clarity Insights acquisition reinforces Accenture’s strategy around AI and life sciences that the INTIENT purchase supported. The report adds, “TBR believes Accenture must foster industry-specific partnerships to extend the capabilities of INTIENT and drive traction for the platform in the industry.” TBR will closely track how Accenture’s partnerships evolve and how the company drives new revenue based on these acquisitions.

Echoing Accenture’s focus on AI, Amazon acquired Health Navigator, a platform designed to foster more expeditious collaboration between healthcare providers and patients, in part through natural language processing and enhanced analytics. Amazon reportedly purchased the company amid efforts to build out Amazon Care, its in-house healthcare services, which it launched in September 2019. On the surface, Amazon’s healthcare-related acquisitions and moves denote neither an immediate threat to traditional HITS vendors nor a clear signal Amazon intends to become a different kind of player in the HITS space. Analyzing Amazon only on the surface would be foolishly shortsighted. Once the company irons out the challenges within Amazon Care, including fully integrating Health Navigator, TBR expects the company will craft a new offering for Amazon clients, potentially starting first with healthcare joint venture partners JPMorgan (NYSE: JPM) and Berkshire Hathaway (NYSE: BRK.A; NYSE: BRK.B). At 1.2 million employees for those three companies combined, Amazon would have a sizable test bed for enhancing current capabilities and developing new offerings. If Amazon can demonstrate an ability to provide top-notch healthcare services for its own employees and a few select partners, every household will wonder if the first step in getting healthcare should start with, “Alexa …”     

In acquiring Fitbit, Alphabet (Google) alarmed some data privacy and industry analysts concerned that the search engine and advertising giant bought the wearables company to gain access to massive amounts of personal, and specifically healthcare-related, data. Both companies’ executives declared data protections would be unchanged and the underlying reasons for the acquisition centered on Fitbit’s expertise and intellectual property around wearable devices and health-tracking applications, platforms and user experience. In TBR’s view, acquiring Fitbit conforms with Google’s overall expansion strategy and specifically boosts the company’s potential role in the overall HITS space. Enhancing Fitbit’s platform with Google’s AI capabilities could further minimize perennial HITS challenges, such as around data privacy and population health, but only if Google can manage the delicate tasks of leveraging user data without violating privacy, crafting and enhancing algorithms that improve the user experience, and maintaining the streamlined seamless flexibility of Fitbit even as the data flows into the highly regulated healthcare ecosystem.  

Quick Quantum Quips: A call for quantum supremacy sends ripples through the market

The quantum market changes rapidly, and the hype can often distract from the realities of the technological developments. In our new monthly newsletter, Quick Quantum Quips (Q3), TBR will brief readers on the latest market announcements, stripping that hype to dig deeper into how recent events will impact the market as a whole. To schedule a time to chat with Analyst Stephanie Long or another one of TBR’s quantum analysts about any of the insights below, contact her at [email protected].

October 2019 developments:

  1. Google claimed it achieved quantum supremacy in mid-October, sending ripples through the quantum community. Quantum supremacy is a key milestone many leaders in the quantum computing space have been working toward for years. If true, this milestone would mean that quantum theory has successfully been translated into practical applications, so such a claim has major implications for the industry overall. Google claims its quantum computer was able to perform a truly random number generation in 200 seconds — and that the task would have taken a supercomputer 10,000 years to complete. Further, truly random number generation is necessary for quantum-safe security solutions, making this announcement a multifaceted milestone in the quantum community. Critics of Google’s claim state that it is possible to achieve very similar results in 2.5 days on a supercomputer, although it would require 250 petabytes of storage to do so, potentially diminishing the size of Google’s “milestone” achievement but confirming it as an achievement nonetheless.
  • IBM has been in the news consistently during October for its strong claims against Google’s quantum supremacy claims. TBR believes that the strong opposition signifies the power being the first company to achieve quantum supremacy can hold as well as the damage to the industry an unrealistic claim can cause through false hyping of the technology. The industry already struggles with hype, which pushes C-Suite executives to invest in and expect quick results from a technology that is meant for the long game, and skewed claims only stand to increase the negative impacts of the hype. As such, IBM has made a significant effort to minimize the hype surrounding Google’s announcement to reveal the complete facts surrounding the achievement.
  • IonQ received its latest round of funding in October — to the tune of $55 million. Samsung and a sovereign wealth fund of the United Arab Emirates led the funding this round, while Google, Amazon and New Enterprise Associates re-upped their commitments from earlier funding rounds. The investments in IonQ are significant, as the list includes some potential competitors such as Google. TBR notes that Google is investing in superconducting quantum computing, which presently leads the charge in terms of advancements. However, IonQ’s theory of trapped ion quantum computing is unique in that it does not require cryogenically cold environments to function, making its approach seem more realistic in that it would have broader, more practical commercial applicability. TBR believes Google’s investment in IonQ demonstrates its strong cash position and focus on the applied uses for quantum over being wedded to any particular hardware structure. Google, like many enterprises, is more focused on application exploration rather than the sale of quantum systems.

  • In a true demonstration of the sheer power quantum computing can unleash, customers are jumping on the innovation train to accelerate the development of both the technology and related skills. Airbus announced that it has compiled a list of leading experts to act as judges for its quantum computing competition. The Airbus Quantum Computing Challenge launched earlier this year and is designed to encourage experts and those interested in quantum computing to tackle some of the more complex computational problems for aerospace. All proposals needed to be received by Oct. 31 and are now being reviewed by the team of judges that Airbus compiled. Jury members come from all geographies and from both industrial and academic organizations, including QC Ware, Horizons Quantum Computing, the University of Waterloo, the University of Technology Sydney, QuSoft, the University of California and more. The announcement is significant because a commercial enterprise is recognizing the value quantum can bring to its business and displaying an eagerness to contribute to the advancement of the quantum ecosystem.
  • QC Ware unveiled the list of speakers for its upcoming quantum computing event, Q2B. The event will take place in California in December. Program details can be found at the link provided.

If you would like your company’s announcement featured in an upcoming Q3, contact Geoff Woollacott to coordinate a conversation.

Embedding multicloud and software-driven services in portfolios helps vendors execute on strategy, expand addressable markets

Google Cloud revenue surpassed the $2 billion mark in 2Q19, doubling in size in six quarters. Under the guidance of CEO Thomas Kurian, Google Cloud is improving its enterprise appeal by launching its multicloud management tool set, Anthos; leveraging acquisitions to build out its migration, storage and analytics capabilities; and expanding its global sales and delivery capacity. Similarly, Salesforce complements internal innovation around solutions such as Customer 360 with ongoing acquisition activity and investment in its partner network. TBR estimates the vendor attained $3.95 billion in revenue as sales teams expanded single-product customer engagements, many of which are led by Service Cloud, into multiproduct deals.

Additional assessments publishing this week from our analyst teams

Capgemini continues to gain momentum in cloud services, with cloud revenue driven by offerings in the Capgemini Cloud Platform portfolio, which supports clients when building, migrating and managing applications and infrastructures in cloud environments. By delivering a cloud-first option, Capgemini enables enterprise and public sector clients to become agile through offerings related to data center modernization, cloud-native solutions, application modernization, intelligent applications, and emerging technologies such as IoT, blockchain and AI. Offering each client its entire portfolio enables Capgemini to provide holistic transformational solutions and effectively compete with peers. Elitsa Bakalova, Senior Analyst

TBR’s Public Sector IT Services Research practice will publish its 2Q19 ManTech report this week.  With top-line revenue expanding 9.4% year-to-year to $537 million, ManTech should be one of the top-performing vendors in 2Q19 in terms of sales growth. ManTech’s top-line expansion owes largely to accelerating spend among classified customers in the Department of Defense (DOD) and Intelligence Community that are increasingly engaging ManTech to enhance warfighting capabilities across all domains, but particularly in space and cyber. ManTech’s addressable market is set to expand and diversify into the civilian sector as the integration of Kforce Government Solutions (KGS) continues. KGS will add 500 employees with large-scale IT infrastructure modernization and transformation expertise, primarily with the Department of Veterans Affairs, to ManTech, while contributing roughly $100 million in revenue (based on revenue of $98 million reported by KGS in 2018). Inorganic sales will largely accrue in ManTech’s Mission Solutions and Services segment, where the core customer focus is the DOD, the Department of Homeland Security and federal health agencies. Look for TBR’s 2Q19 Perspecta report next week, as we examine how the company is leveraging its R&D-led approach to maintain its growth momentum as it begins its second full year as an independent, federal IT competitor. John Caucis, Senior Analyst

Weakness in Cognizant’s core industry segments overshadowed increased growth in digital in 2Q19. The company’s ability to rapidly scale its digital revenue will be key to offsetting this weakness, specifically in Financial Services and Healthcare. In the near term, Cognizant must emphasize cross-sales of acquired assets, such as Zenith Technologies, within its existing and acquired install bases. Kelly Lesiczka, Analyst

Salesforce to acquire Tableau as the market moves to no-code and low-code environments

Salesforce’s acquisition of Tableau Software would make additional data visualization and analytics capabilities available to business users

Salesforce announced its intent to buy Tableau Software in an all-stock deal valued at $15.7 billion, in which Salesforce will exchange 1.103 shares of its common stock for each of Tableau’s Class A and Class B common stock. Salesforce has been improving its Einstein analytics capabilities with functionality such as preconfigured templates and drag-and-drop analytics that enables users to create data visualizations without code. Tableau would expand Salesforce’s data visualization and analytics capabilities, as its data sorting and no-code and low-code data visualization capabilities will enable business users to manipulate data and create new data visualizations without a data scientist. While Tableau would continue to run as an independent company, TBR expects Salesforce would create integrations between Tableau and Customer 360, Salesforce’s app that connects data from its sales, service, marketing and e-commerce offerings. Since data from Customer 360 can also be pulled into Einstein analytics, the integration of Customer 360 with Tableau would enable Salesforce customers to leverage Tableau’s technology from a central touch point. If finalized, the acquisition would also increase Salesforce’s value proposition as a front-office provider for Tableau’s 86,000 customers, creating new cross-selling opportunities.

Salesforce’s front-office provider partners and competitors develop no-code- and low-code-enabled data visualization analytics as the industry trends toward codeless data manipulation

Salesforce’s announcement comes four days after Google announced its intent to acquire Looker, an analytics and data visualization company. While it is unlikely that Salesforce’s planned acquisition of Tableau was in reaction to Google, these developments further highlight the increasing importance of data visualization in the public cloud market. TBR does not expect Salesforce’s partnership with Google would be significantly impacted by the vendors’ recent investments, as Salesforce’s broader SaaS portfolio still targets a larger front-office audience and Google remains a strong IaaS partner. However, Salesforce’s acquisition of Tableau would improve Salesforce’s position against Oracle in the front-office space. Oracle is improving its own analytics capabilities and unifying its data through CX Unity, its offering akin to Salesforce’s Customer 360. Additionally, the same day Salesforce announced its intent with Tableau, Microsoft announced AI Builder, which makes it easier for its PowerApps and Microsoft Flow customers to manipulate data to create AI models. Microsoft’s Power BI is in the same Power Platform product family as PowerApps and Microsoft Flow, showing Microsoft is trying to make it easier for customers to manipulate data as well. However, Microsoft’s Power Platform solutions still require a layer of coding and technical knowledge that exceeds the skill set of the typical business user. Given Salesforce’s and Google’s announcements, TBR expects Salesforce competitors such as Oracle and Microsoft may acquire data visualization companies to quickly enhance their capabilities in the space and level the balance of power in the public cloud front-office market.

Webscale capex growth will decelerate, though dollar volume will continue to climb, as data center builds slow

According to Technology Business Research, Inc.’s (TBR) 1Q19 Webscale ICT Market Landscape, webscale ICT capex for the Super 7 will grow at an 8.1% CAGR to nearly $58 billion in 2023. Most U.S.- and China-based webscales began pulling forward significant investment in data center and network capacity in 2018, which will lead to moderating — or even declining — capex levels for some U.S.-based players beginning in 2020. China-based webscales will continue to ramp ICT capex through the forecast period, however, to catch up to Western rivals in key areas, particularly public cloud.

Webscale "Super 7" capex forecast

The entrance of Rakuten, a Japan-based e-commerce company, to the mobile industry could be a game changer and provides a glimpse into what a digital service provider will look like. Rakuten’s mobile network will blanket Japan with LTE coverage by year-end. Not only will Rakuten’s network be agile, flexible and dynamic to provide digital services, it will also enable a dramatic reduction in the cost of connectivity. Rakuten’s ultimate intention is to be more than just another mobile network operator in the highly competitive Japan market; it aims to provide a foundational connectivity platform from which to sell a host of digital services. Rakuten’s acknowledgment that it needs its own network could lead to other webscales trying to take a more active ownership and control stance toward having a connectivity platform from which they can leverage their digital businesses. Alphabet, Facebook and Amazon, among other webscales, have all experimented with how to address last-mile connectivity, not only to bridge the digital divide but also to serve as a conduit to give them more control over their destinies without relying on communication service providers (CSPs) to provide the connectivity layer.

The OEM landscape continues to see disruption due in part to the power webscales hold over their suppliers. The vast number of suppliers taking part in Rakuten’s network build demonstrates that webscales hold the power when soliciting vendors for connectivity initiatives. When engaging with webscales, which have few legacy encumbrances, incumbent OEMs are being relegated to commoditized hardware and services. Should the 5G era bring about this trend in the CSP customer segment, incumbents will see more widespread disruption. Vendors must be wary of the webscale procurement model taking hold with their traditional customers.

Cost of ‘intelligent connectivity’ must decline significantly for intelligent world to unfold

TBR perspective

Realizing the intelligent world presented by the mobile industry at Mobile World Congress Barcelona 2019 (MWC19) will require a fundamental change in how networks are architected, including a radical reduction in the cost of providing connectivity. It will also require business transformation for companies tied to the old world, namely communications service providers (CSPs) and their incumbent vendors.

It was readily apparent at the event that technology is advancing at a much faster pace than the establishment of business cases that economically justify deployment of the technology. The reality for the mobile industry is that the cost of building, owning and operating networks is too high and networks are too inflexible to support the business realities of the digital era, whereby connectivity is relegated to a commodity service and the value lies in the platforms and applications that run over the network. The industry has known this for years, but changes have been minimal, until maybe now.

The entrance of Rakuten to the mobile industry could be a game changer and provides a glimpse into what a digital service provider will look like. In what could arguably be the most important takeaway from the entire event, Rakuten’s approach to building and operating a network could signify a paradigm shift in the industry. Not only will Rakuten’s network be agile, flexible and dynamic to provide digital services, it will also enable a dramatic reduction in the cost of connectivity.

The theme of MWC19 was “intelligent connectivity” and centered on how 5G, IoT, AI and big data are coming together to enable the intelligent world. Against this backdrop, Rakuten stole the show with the evangelization of its end-to-end virtualized and cloud-native network, which is being deployed across Japan this year. Rakuten’s network provides a glimpse into what the intelligent network of the future will look like.

AI chips: Explosive growth of deep learning is leading to rapid evolution of diverse, dedicated processors

Artificial intelligence (AI) utilization has been accelerating rapidly for more than 10 years, as decreases in memory, storage and computation cost have made an increasing number of applications cost-effective. The technique of deep learning has emerged as the most useful. Large public websites such as Facebook (Nasdaq: FB) and Amazon (Nasdaq: AMZN), with enormous stores of data on user behavior and a clear benefit from influencing user behavior, were among the earliest adopters and continue to expand such techniques. Publicly visible applications include speech recognition, natural language processing and image recognition. Other high-value applications include network threat detection, credit fraud detection and pharmaceutical research.

Deep learning techniques are based on neural networks, inspired by animal brain structure. Neural networks perform successive computations on large amounts of data. Each iteration operates on the results of the prior computation, which is why the process is called “deep.” Deep learning relies on large amounts computation. In fact, deep learning techniques are well known; the recent growth is driven by decreasing costs of data acquisition, data transmission, data storage and computation. The new processors all aim to lower the cost of computation.

The new chips are less costly than CPUs for running deep learning workloads

Each computation is limited and tends to require relatively low precision, necessitating fewer bits than found in typical CPU operations. Deep learning computations are mostly tensor operations — predominantly matrix multiplication — and parallel tensor processing is the heart of many specialized AI chips. Traditional CPUs are relatively inefficient in carrying out this kind of processing. They cannot process many operations at the same time, and they deliver precision and capacity for complex computations that are not needed.

Nvidia (Nasdaq: NVDA) GPUs led the wave of new processors. In 2012, Google announced that its Google Brain deep learning project to recognize images of cats was powered by Nvidia GPUs, resulting in a hundredfold improvement in performance over conventional CPUs. With this kind of endorsement and with the widespread acceptance of the importance of deep learning, many companies, large and small, are following the money and investing in new types of processors. It is not certain that the GPU will be a long-term winner; successful applications of FPGAs and TPUs are plentiful.