Google is at a very delicate moment.
Despite the change in the U.S. government, political alarms remain in place against the largest U.S. Internet information and software technology company. In the United States alone, Google is facing at least three antitrust investigations at the federal level and in the vast majority of U.S. states.
The way Google Inc. is organized and run is also undergoing a major test, and Sundar Pichai has not had a good track record since he took over the company in 2015.
Now is the time when Google needs to strongly integrate its diverse and varied businesses. However, Pichai’s financial results since he took office show that Google is still a company where the core search/advertising business accounts for the absolute majority of the company’s revenue, while the businesses outside the core still haven’t made much progress and the company faces multiple management challenges.
Google’s strong AI research force, the output of which has been partially integrated into the business. But despite Google’s claims that the world has shifted to an “AI-first” world, the overall level of integration between its AI capabilities and its business is still low, and it can’t shake the revenue position of its core business. As for companies like DeepMind, which continues to expand the boundaries of AI technology, they are still unable to find a north in commercialization.
as one of the first batch of dedicated AI hardware announced in Silicon Valley, Google’s TPU has lagged behind in terms of openness to external users from the beginning and has so far failed to be accessible to more AI researchers, lagging behind major competitors such as Nvidia in terms of voice and sales.
The many “Other Bets” projects spun off from Google in 2015, only a few of which have survived so far; Waymo, the earliest and loudest self-driving company, has made little progress in commercial advancement and its valuation has shrunk significantly.
Relying on its own control of the operating system ecology, Google’s Pixel cell phone business has lasted for many years, making some competitive products, but after spending the same or even more time and financial resources as some fellow startups, it still can’t stand firm in the industry and is not well received.
Google Cloud, which has become an independent company, has seen good revenue growth in the past two years, but there have been many internal and external complaints: its top-to-bottom management model is considered disconnected from Google, and there are few employees who want to transfer back to headquarters; external users cannot find AWS and Azure on Google Cloud, which gives priority to customer service.
Google continues to expand its workforce, with more than 135,000 full-time employees (plus a sizeable outsourced workforce) by the end of fiscal 2020. With high salaries, enjoying a rather generous package in Silicon Valley and around the world, many of these employees are complaining about the company’s actions and inactions in diversity, government business, corporate politics, etc. ……
This inevitably makes today’s Google, look a bit like the former Baidu: founded many years later still overly dependent on the core business, can not keep up with the trend of the times, many transformation attempts to end without success.
AI: Starting early, catching a late start
Google was the first Silicon Valley company in its time to make cutting-edge technology research a corporate development priority, and probably not one of them. Under the development and leadership of Jeff Dean, Google applied distributed computing systems to the company’s advertising, web crawling, indexing and other systems, and established a research department dedicated to machine learning/deep learning early on. Google Brain).
In 2017, Pichai posted that the world was undergoing a major tech transformation, moving from mobile-first to AI-first.
Google’s years of sustained investment in scientific research is making pretty good progress today. Papers from Google researchers continue to lead other companies in terms of the number of historical citations. And at the application level, many of Google’s core business products today have been augmented by AI research.
The core algorithms of Google’s search engine have integrated deep learning technologies to generate more efficient and useful search results for users; Google Ads and Doubleclick products have adopted Smart Bidding, an intelligent automated bidding technology based on machine learning technologies; and many other products, including Maps, Gmail, Translate, Chrome, Photos, and many others, have been enhanced by AI research. Chrome, Photos, and many other products have also benefited from the results of Google AI research.
For example, when Amit Singhal, former head of Google Search, left in 2016, his successor, John Giannandrea, was one of the key leaders of Google’s AI research, driving the progress of integrating AI research in the search business.
But that’s only one side of the story. Pichai is right that the world has indeed become an AI-first world, but it seems that Google itself is facing a disconnect between its undeniable leadership in AI research and the company’s revenue growth. AI is used in search, but Google is still a search-first company.
In Google AI’s annual report, the research-business integration is short and substantive content is often rarely mentioned. Google’s AI technology is making good progress in integrating with some consumer-facing products, but most of these products, except for those based on Google’s integrated user account system, which can provide more data for the core business and build a more comprehensive user profile, cannot directly bring in revenue.
Moreover, some of these products, such as Google Hangout, which has chosen the direction of office software after several transformations, are also facing a strong squeeze from competitors such as Slack. Similar situations cast another shadow over the future of Google AI research applied to products to create business value and continue to maintain the company’s industry influence.
Google’s track record in AI hardware is also worth mentioning.
Google developed the TPU (Tensor Processing Unit), an early batch of specialized computing devices for deep learning training and inference in Silicon Valley, which received much attention at launch. At that time, the Google-led TensorFlow deep learning framework was popular among researchers and developers for its high performance, wide adaptability, and low barrier to entry, and the TPU had the promise of further expanding Google’s strengths in deep learning from the algorithm to the hardware level, achieving a true combination of hardware and software. At the time, some bullish people even compared Google to the Intel of the new era of AI computing.
A few years have passed, and the advancement of TPU has not been what people expected. As the product iterated, TPU and Cloud TPU grew exponentially in terms of computing power on paper. However, for some time after its release, the hardware has been limited to internal use and very limited third-party use applications due to TPU’s computing environment requirements and Google’s baffling strategy, and has not been made available to a wide range of developers as soon as possible.
For now, third-party developers need to become Google Cloud users, or purchase the Edge TPU, which was only released in 2018 and has much less computing power and is far less eye-catching than the “genuine” TPU, (or through the Pixel4 phone, which includes some of the Edge TPU’s features) — to be able to deal with TPUs. However, these developers also have GPU-based options such as AWS and Microsoft Azure to choose from.
This situation has objectively limited the adoption of TPUs on the ground in the industry in the early years of its release.
Silicon Valley chip-based companies, represented by Nvidia, have matched or even surpassed Google’s progress and have continued to release GPUs, CPUs, FPGAs or ASICs specifically for deep learning in recent years. In the early years when Google released TPU to compare it with Nvidia inter-generational shelf GPU architecture, it also slightly makes people wonder whether Google is not “honest” enough in technology.
Non-core business: few and far between
In 2015, Google completed an independent split and reorganization, with its core businesses continuing to operate under the Google brand; non-core businesses, such as venture fund GV, fiber broadband Fiber, self-driving company Waymo, healthcare Verily, drone Wing, and moon factory (formerly Google X Labs), were largely independent. The company’s new parent company, Alphabet, is overseen by “Other Bets” (other bets).
(Pichai was also named Alphabet’s CEO in 2019.)
Five years after the reorganization, the vast majority of Other Bets’ projects are operating haphazardly, with dismal results, and still rely heavily on Google as a backstop, with some shutting down. The very few high-profile projects that have survived, even the “Google-based” companies that were able to secure high funding, are not in an absolutely safe position.
Most projects have stalled. Fiber broadband service is currently slow to roll out across the United States, only a limited number of urban households can apply, and coverage does not exist in a regional sense, only a spot presence.
Google Fiber coverage in Seattle
Then there is Verily’s lack of independence, having previously borrowed Google offices for quite some time, sharing shuttle cafeterias and internal recruiting tools, etc. (Verily even once had a dispute with Google’s headquarters over costs). When the new epidemic arrived, Verily quickly developed a preliminary medical screening website, but the team and business were so immature that the functionality was basically the same as a questionnaire written by a random third party, and the results were not good; it was not until six months after the outbreak that Verily developed a return-to-work review tool that it got its first customer, sister company Waymo. -sister company Waymo.
Before that, because of the existence of the cash cow search advertising business, there were still pockets of these projects. However, one of the key reasons for Google’s restructuring in 2015 was that a significant portion of employees, executives, and shareholders believed that these money-burning projects with no end in sight were a drag on Google’s overall financial performance. The act of breaking up and restructuring is itself a wake-up call for these companies. One of the key responsibilities of Google and Alphabet CFO Ruth Porat is to oversee that they abandon their illusions and make money as soon as possible.
Yet five years after the reorganization, a large part of Other Bets projects (mainly those running physical businesses) are still haphazardly commercialized, with no clue. Google’s Other Bets projects reported a total loss of $4.5 billion in fiscal 2020 – an improvement over the previous year, but still a significant loss problem, according to the report. Since the reorganization, a few projects have been forced to shut down (like Project Loon) and some have been reunited with Google (like Jigsaw), and there are still about ten companies under Alphabet that fall under the Other Bets category.
Waymo is a prime example.
Waymo, which was born out of Google’s self-driving technology team and was one of the first self-driving projects to be launched from within a major company in Silicon Valley, first attracted a lot of attention years ago when it released its first self-driving car. The model was sleek and cute, and also featured a forward-thinking design concept of a driver-free interface.
While Waymo once racked up the most self-driving test miles of any self-driving company/program in the country, and also launched a public urban road testing program in 2017, once ahead of its main competitors in terms of self-driving technology – behind the scenery, Waymo has been very slow in business development.
There are reports that Waymo has cooperated with FCA, Ford and other mainstream car companies, but could not reach a consensus on the development plan, and eventually the cooperation failed to take shape; within Alphabet, Waymo’s R&D costs were too high, and the parent company had to urge Waymo to finance from outside.
The recent announcement of Waymo CEO John Krafcik’s departure to Aurora, a company founded by rival and former Google self-driving team founder Chris Urmson, is widely believed to be the root cause of the difficult relationship between Waymo and Alphabet. In general, Alphabet is very unhappy with the state of Waymo’s operations due to the slow progress of the business.
Another rather extreme example might be DeepMind.
The AlphaGo Go AI developed by this UK-based company has defeated the world’s top-ranked players, Lee Se-dol and Ko Jie, giving a first glimpse of the power of AI technology to a more tech-savvy public.
Behind the hullabaloo, the company has been making serious losses for years: $470 million in 2018 and £477 million in 2019. The losses come mainly from payroll: based on 2018 figures, the company’s staff salaries cost up to $580,000 each.
The situation was similar last year, with the bulk of the losses still coming from human resources, with DeepMind’s total number of employees increasing to 1,000. To ease the financial pressure, DeepMind’s donor support to academia shrank from £13.5 million in 2018 to £6.3 million in 2019, even as Google Ireland had to forgive a debt of up to £1.1 billion.
The expensive human cost has contributed to DeepMind’s leadership in AI research. It has developed unsupervised learning-related machine learning technologies such as AlphaGo and AlphaFold, a genetic protein prediction technology, in the past few years, which is unique and unmatched in the related field.
However, years after being acquired by Google and integrated into the Google family, this technological sophistication still struggles to be translated into healthy revenue. 2018 saw DeepMind’s $125 million in revenue come entirely from YouTube, Waymo, Google Cloud, and other Alphabet “Google family ” connections.
It’s clear that today DeepMind is in deep commercial trouble. The fact that Google is allowing DeepMind to burn money (or even burning money for it) is tantamount to forcibly renewing its life.
Google Cloud: promising growth, cultural disconnect
When cloud computing company Bebop was acquired by Google, its founder, Diane Greene, already had a good idea of Google’s ambitions for the cloud business. After three years as a Google board member, Greene was named CEO of Google Cloud.
In terms of branding and structure, Google Cloud is still a unit under Google, but the CEO position is unprecedented at Google. However, after a period of time in office Greene found that his autonomy was not that high.
But the development of Google Cloud, in general, is still relatively smooth. early 2019 when Greene left the post, Google Cloud reached $5.8 billion in revenue in fiscal year 2018, and the cumulative annualized revenue during his tenure has been $8 billion, which is an excellent achievement for a latecomer in the cloud computing field.
Thomas Kurian, the new CEO of Google Cloud, enjoys an even higher level of autonomy than Greene. To sustain business growth, he has largely changed the way the company is organized and managed and the corporate culture. Compared to Google’s headquarters, where an engineer culture prevails and much innovation benefits from a bottom-up approach, Google Cloud’s atmosphere is less liberal and inclusive, and the management model is top-down.
The model is commercially successful, with Google Cloud revenues of $13 billion in fiscal 2020, still far behind Microsoft Azure’s $50 billion and AWS’s $45 billion, but meeting market expectations.
Google Cloud’s headquarters building in Sunnyvale
At the same time, Google Cloud has become increasingly controversial, both internally and externally.
In early 2020, the media reported that Google Cloud was planning to start a series of layoffs at its Silicon Valley headquarters. No one wants to lose their job, let alone their job at a Google-affiliated company, but before then, a number of Google Cloud employees had already privately expressed their dissatisfaction with the company’s top-down management model, and even wanted to be internally transferred back to Google’s headquarters.
In the view of these employees, Google Cloud’s management style has made the Google-based cloud computing company lose the look of the Google system. These employees, many of whom previously worked in Google’s consumer products and other teams, only to become Google Cloud employees as the organization restructured and team reporting changed – feel uncomfortable with Google Cloud’s more old-school enterprise style of management.
One Google Cloud employee had this to say about the proprietor on Glassdoor
Sadly, all the cool stuff you hear about Google doesn’t apply to Google Cloud. There are hordes of self-important managers here to manage you in meaningless detail on a daily basis ……
Teams of sales and sales engineers spend a lot of time focusing on the details within the organization, while spending less time face-to-face with customers …… Many enterprise customers really can’t find any reason to go to Google Cloud, and even Google’s own sales engineers can’t answer the question “Why Google Cloud? ” The question ……
Kurian’s strategy is to recruit more sales to focus on what makes money, but Google is already years behind AWS and Azure in those directions. Such a strategy lacks creativity and foresight ……
On the outside, Google Cloud users generally feel that the company, which is 100% behind Google’s “AI first” strategy, lacks the customer service first experience that AWS and Azure have.
In this regard, Google Cloud does have the air of an old-school enterprise company. And that quick, approachable customer service posture that the Department has cultivated over the years in dealing with the average consumer has not been inherited by Google Cloud. An online search for Google Cloud customer support keywords yielded mostly negative reviews.
One user had this to say about Google Cloud: great technology, crummy customer service: “It’s so good that I’m willing to put up with the pain of customer service and keep using it” – except that in the long run, I wonder how many such loyal Google Cloud customers will remain. customers.
Corporate Culture: Serious Tightening and Conflict
The dilemma of corporate culture and the company’s management model exists not only in Google Cloud, but also along with the objective situation that Google’s business size and employee team are growing wildly, spreading throughout the company and causing more problems.
As we reported exclusively during the new coronavirus epidemic last year, there was a group of Chinese employees who were active internally, writing letters to company executives and calling their colleagues to action around the accurate translation of the new coronavirus in Google’s different Chinese-speaking markets. The issue in question was highly voted on Google’s intranet, but was deliberately ignored at the company’s TGIF plenary meeting. Fortunately, in the end, these Chinese employees largely achieved their goals.
This incident is just one of the footnotes of how Google’s corporate culture has changed for the worse in the eyes of its employees. In general, the current Google and the old Google, which advocated freedom, openness and flat management, have also produced a considerable gap.
The full name of TGIF is “Thank God it’s Friday”, which means that the weekly plenary meeting on Friday has been changed to once a month. Questions are limited to those related to business, and executives are no longer able to speak freely.
Google still requires employees to work “like a startup” out of a desire for revenue, speed, and efficiency (as do many other large Silicon Valley companies of the same size). But when it comes to the team, employees find that many confidential information is restricted, and managers/teams are fighting for resources, so they don’t get the startup experience of being united as a team.
At the same time, conflicts between employees and the company began to emerge. In recent years, Google has become the home of social movements in Silicon Valley.
A white paper by James Damore attacking female employees as less skilled than male employees sparked an internal outcry; in the middle, tens of thousands of employees organized a strike to question why the company offered an outrageous exit package to executives suspected of sexual harassment; later, thousands of employees signed an open letter demanding that the company stop its military contracts with the U.S. Department of Defense; and most recently, Google was accused because a member of the company’s AI The recent allegations that Google terminated a member of the company’s AI ethics committee because his ideas were contrary to the company’s interests have once again led to protests by many employees and external AI experts.
To be clear: these employee protests are not all really about what Google did wrong.
Rather, the existence of these protests increasingly highlights the intensification of employee grievances and the problems with the dialogue mechanism between employees and the company. The legitimate parts of their claims are not being dealt with in an equivalent and reasonable way under the company’s current operating model.
It goes without saying that big companies are bound to encounter such problems. But at least for Google, Pichai’s rise to the top has not brought a proper solution to these problems. Pichai is a reassuring CEO for the board of directors, but not a good “boss” who can represent them and do justice for them among a significant part of the employees.
In any case, Pichai is already at the top of the hierarchy at Google, and at Google as a whole. Google founders Larry Page and Sergei Brin, as well as one-time guardian Eric Schmidt, have largely stayed away from the company’s troubles. Page himself is distressed by the global strike of Google employees and fears that all their efforts over the years have been in vain, The Economist reported, citing people familiar with the matter.
But at this point, they can only continue to trust and rely on Pichai to take Google out of this quagmire.
Today, Google remains the best face of Silicon Valley’s innovative spirit. Its R&D efforts in machine learning, cloud computing, and many other areas continue to serve a wide range of users, and its investment in quantum computing promises to pave the way and bridge the way for future technological advances.
What makes Google employees most proud is that they can develop and maintain a wide variety of products every day, serving billions of users, and changing the world with the power of technology in the process – even though much of what they do is not profitable.
In other areas not yet mentioned in this article, Google has a lot of problems of one sort or another. For example, the company is notoriously keen to “feed the compulsion” by developing a lot of competing internal projects of the same kind, leading to serious internal friction, demotivating employees to innovate and then, unsurprisingly, killing those projects.
In addition, Google’s once well-known “20% time” policy, which allowed employees to spend one day a week working on individual projects (not mandatorily related to company development), was eliminated in 2013 and replaced by an internal incubator called Area120 in recent years. — both of which, while both encourage employees to try new things outside of their main business and during work hours, have very different incentive effects on employee innovation.
What’s more, Google’s core businesses are now under considerable threat. On the one hand, Alphabet’s own revenue growth (mainly from search advertising) has declined this year; on the other hand, Mark Shmulik, vice president of Bernstein, a market research firm, estimates that Amazon now controls 60% of product-related searches worldwide, while Amazon’s advertising business has increased significantly in recent years.
Google is a public company, and all the aforementioned issues of business integration, commercialization of cutting-edge technologies, corporate culture and labor conflicts are bound to delay its healthy development if they persist.
In the current anti-trust investigation against Silicon Valley companies, there is a voice in the community of supporters: splitting up big tech.
Most tech workers probably don’t want to see Google split up. The interesting and “romantic” thing about this most representative Silicon Valley company is that it does everything well, so what’s the point of breaking it up?
Google itself must not want to be forced to split up. However, if the problems of new businesses and new technologies, and the “difficult integration” between businesses and businesses continue, a more awkward situation may arise: even if Google is split up, it will not have much impact and meaning.
In the eyes of Google’s supporters, the company is today very successful, profitable, and has a top-notch team of employees and excellent research and development – with these elements, the above-mentioned problems should not be a problem.
But Silicon Valley has no shortage of companies with the same elements, such as Hewlett-Packard, which has gradually fallen behind and been forgotten in history.
Recent Comments