Google's SGE within Search Labs is beginning to offer source links for queries in Search.
It's not as widespread, though, it has displayed several source links after pieces of information in case users would like to double check.
Google has wanted to ensure it is developing "responsible" AI and its source linking was also seen mentioned for its AI NotebookLM tool.
It looks like Google is beginning to test a way of bringing a little more reliability to its AI-powered Search experience.
According to 9to5Google, the company has started a test for its SGE (Search Generative Experience) where it displays very clear in-line source links. Though it's not quite as widespread, it looks like the source links will offer the name of the publication or website directly after a statement so users can do some fact-checking.
The first instance of these links appearing showed Google citing three websites for a piece of information it offered on the Galaxy Z Flip 5.
The AI software continued this trend throughout all of its provided information, with article previews in a carousel on the right side of the text box if users would prefer to look there.
The bot also offered some follow-up questions to the 9to5's query.
Google's SGE was detailed a bit after its I/O event back in May. The company opened access to Search Labs so users could experiment and test its generative AI for its search engine. The purpose of it was to deliver quick summaries of relevant information about a user's query so they could get caught up quickly without needing to do a lot of excessive reading.
It was also during this time Google showed off its generative AI's existence in search through its colored text boxes alongside the ability to ask a follow-up question.
Moreover, Google's inclusion of more prominent source linking is probably due to its interest in creating "responsible" AI mechanics and tools for users. We've seen the company show an interest in ensuring users can trust its AI when it announced NotebookLM.
Google's AI notetaking app was made to help students and others gather facts from their notes quickly. Notetakers will also gain an AI bot tailored specifically for whatever Topic they're interested in to help source more information from the web.
"Source" is key here at NotebookLM's AI software will also provide a link to where it took the information from so users can fact-check its work.
Once a miniature playground for the public to test Google’s blossoming AI-powered tools, the AI Test Kitchen app has been delisted for both Android and iOS, moving solely to the web.
Update: Google has shared additional details about the shutdown of these mobile apps.
When it first launched last August, AI Test Kitchen was home to three initial experimental demos with the promise of more to come.
AI Test Kitchen will be home to a “rotating set of experimental demos” and three are currently available. They are based on the “latest version of LaMDA, which has undergone key safety improvements.”
- Imagine It: Name a place and LaMDA will offer paths to explore your imagination.
- List It: Name a goal or Topic and see how much LaMDA can break it down into multiple lists of subtasks.
- Talk About It (Dogs Edition): Roll with the conversation and see where it goes. It’s just a fun, kinda-weird, open-ended chat.
While Google announced that a “Season 2” for the app would be coming, even detailing what new tools would appear, it seems those plans were scrapped at some point. Instead, AI Test Kitchen gained MusicLM at this past Google I/O and simultaneously dropped support for the earlier demos.
Update 8/1: A Google spokesperson has confirmed that while AI Test Kitchen’s mobile apps are shutting down, the playground itself is simply moving to the web. In the hours after the mobile apps were delisted, the apps themselves stopped working, pushing users to use the web experience instead.
The move is intended to allow AI Test Kitchen to be updated more rapidly and easily across all devices instead of pushing separate updates to Android, iOS, and the web.
We’ve updated our coverage to reflect that this latest change only affects the mobile apps.
On Monday, AI Test Kitchen’s mobile apps began winding down, as they’re no longer publicly listed on the Google Play Store or Apple App Store (though they may still be visible to those who previously installed them). Those who previously used the mobile app can continue to access MusicLM (and any potential future additions) through the web app.
Considering Google has always positioned AI Test Kitchen as a place for experiments, these mobile apps shutting down isn’t much of a surprise. That said, it’s still somewhat surprising to see the mobile apps seemingly discontinued less than a year after the initial public launch.
Hopefully, we’ll see Google continue to experiment with delightful AI demos going forward in a more public setting instead of through a dedicated app. At the very least, I hope the surprisingly listenable MusicLM isn’t going anywhere anytime soon.
FTC: We use income earning auto affiliate links. More.
choness
Tech is hot once more. The Magnificent 7 has been red hot in 2023 and helped drive the S&P up 20%.
Ycharts
Most of these gains haven't been justified by fundamentals, not even close.
A few red-hot tech names have had their gains sanctified by the righteousness of fundamentals.
Let me show you why Google (NASDAQ:GOOG) (NASDAQ:GOOGL) is one of my favorite hyper-growth ultra SWANs.
One that could triple in the next five years and almost quadruple the market's returns in 18 months. Why?
Because of almost $500 billion worth of expected buybacks through 2028.
And over $500 billion in cash after half a trillion in buybacks by 2028.
And GOOG is expected to generate $700 billion in free cash flow in the next six years, and the rewards for investors could be 5X better than the S&P 500. (see valuation section)
Google has more money than the Vatican (which has $65 billion) and is a red-hot profit machine that's just getting warmed up.
Seeking Alpha
The market loved Google's results even if they were not exactly impressive.
Total second-quarter revenue was $74.6 billion, up 7% from last year. Advertising revenue returned to growth (up more than 3%) after two consecutive quarters of a decline, with improvements in both searches (up 5.6%) which was driven by strength in retail, and YouTube (up 4.4%), partially offset by the ongoing weakness in advertising technology revenue (down 5%). Cloud revenue increased 28%, and other services, which include hardware and Google Play, were up 10%." - Morningstar
What matters to Google in the long term is the strong growth in the cloud. Remember that cloud is the future of AI and pretty much everything tech.
Most of the value added to the enterprise in the future will be in the cloud.
Statista
GOOG is one of the big three in the cloud, and no one is even close in fourth.
Google Cloud is running $33 billion per year in sales and generating $8.8 billion in annual cash flow.
By 2025 analysts expect Google Cloud to hit $51 billion in sales, but due to high economies of scale, cash flow from the cloud is expected to double to $16 billion.
Today cloud makes up 7% of GOOG's cash flow, and in 2025 it's expected to hit 10%.
Cloud sales growth is expected to grow at 21% annually through 2028, as GOOG wins market share, mostly from the lower tier rivals who can't keep up with the titans.
Google's sales are growing at 10%, and the cloud is growing twice as fast.
Why am I so bullish on GOOG long term?
Because what it's doing with AI has the potential to change the very nature of the global economy and even how humans forever live their lives.
But in the short to medium term, thanks largely to GOOG killing it in the cloud, the growth prospects for one of the world's best companies keep improving.
Short-Term Growth Consensus
Metric | 2022 Growth | 2023 Growth Consensus | 2024 Growth Consensus |
2025 Consensus |
Sales | 12% | 12% | 11% | 10% |
EPS | -19% | 24% | 17% | 16% |
Operating Cash Flow | 3% | 19% | 17% | 11% |
Free Cash Flow | -8% | 24% | 23% | 18% |
EBITDA | 1% | 42% | 14% | 14% |
EBIT (operating income) | -4% | 20% | 15% | 12% |
(Source: FAST Graphs, FactSet)
After a rough 2022, GOOG is expected to recover nicely and grow at a rate even most growth stocks would envy.
FactSet Research Terminal
GOOG's long-term growth consensus has been rising from 14% to 18.2%.
But that's the long-term earnings growth estimate, the median of 53 analysts. What's even more impressive about GOOG is how it's expected to grow all of its fundamentals.
Year | Sales | Free Cash Flow | EBITDA | EBIT (Operating Income) | Net Income |
2022 | $282,836 | $60,010 | $110,132 | $75,483 | $59,972 |
2023 | $303,275 | $74,804 | $121,024 | $87,392 | $71,630 |
2024 | $337,560 | $89,224 | $137,050 | $102,159 | $82,830 |
2025 | $372,796 | $103,210 | $157,663 | $118,106 | $95,210 |
2026 | $413,074 | $118,493 | $173,564 | $129,405 | $103,898 |
2027 | $444,743 | $149,000 | $191,596 | $134,998 | $115,813 |
2028 | $497,230 | $167,819 | $221,772 | $156,718 | $136,835 |
Annualized Growth 2022-2028 | 9.86% | 17.54% | 12.37% | 12.95% | 13.82% |
Cumulative 2023-2027 | $2,368,678 | $702,550 | $1,002,669 | $728,778 | $606,216 |
(Source: FAST Graphs, FactSet)
For a company of GOOG's size to grow free cash flow at almost 20% is extremely impressive.
And guess what GOOG is expected to spend all that cash on?
FactSet Research Terminal
GOOG's cash pile is expected to hit $513 billion by 2028, $313 billion net of debt.
And remember that in 2028 GOOG is expected to generate $167 billion per year of new free cash flow! GOOG is expected to generate $702 billion in free cash flow through 2028.
That's so much cash that even with $482 billion in buybacks in the coming years, its cash pile is still expected to increase by 2000%.
FactSet Research Terminal
Let me repeat that. GOOG is expected to spend half a trillion in buybacks AND still end up with $500 billion in cash.
This company is firing on all cylinders, changing the world in many ways you don't even realize.
GOOG is up almost 50% this year, but it remains a potentially strong buy.
Metric | Historical Fair Value Multiples (all years) | 2022 | 2023 | 2024 | 2025 | 2026 | 12-Month Forward Fair Value |
Earnings | 24.87 | $113.41 | $138.77 | $163.40 | $181.80 | $208.66 | |
Average | $113.41 | $138.77 | $163.40 | $181.80 | $208.66 | $152.98 | |
Current Price | $129.15 | ||||||
Discount To Fair Value |
-13.88% | 6.94% | 20.96% | 28.96% | 38.10% | 15.58% | |
Upside To Fair Value | -12.19% | 7.45% | 26.52% | 40.77% | 61.56% | 18.45% | |
2023 EPS | 2024 EPS | 2023 Weighted EPS | 2024 Weighted EPS | 12-Month Forward PE | 12-Month Average Fair Value Forward PE | Current Forward PE |
Current Forward Cash-Adjusted PE |
$5.58 | $6.57 | $2.36 | $3.79 | $6.15 | 24.9 | 21.0 | 12.2 |
(Source: DK Research Terminal, FactSet)
It trades at just 21X forward earnings, a 16% historical discount. But adjust for its mountain of cash and trading at just 12X earnings.
Right now, private equity is closing deals at about 12X earnings.
Billionaires are getting sweetheart deals at the same valuations that you can buy this AA-rated hyper-growth Ultra SWAN right now.
Rating | Margin Of Safety For Very Low-Risk 13/13 Ultra SWAN | 2023 Fair Value Price | 2024 Fair Value Price | 12-Month Forward Fair Value |
Potentially Reasonable Buy | 0% | $138.77 | $163.40 | $152.98 |
Potentially Good Buy | 5% | $131.84 | $155.23 | $145.33 |
Potentially Strong Buy | 15% | $117.96 | $138.89 | $130.03 |
Potentially Very Strong Buy | 25% | $98.88 | $122.55 | $114.73 |
Potentially Ultra-Value Buy | 35% | $90.20 | $106.21 | $99.44 |
Currently | $129.59 | 6.62% | 20.69% | 15.29% |
Upside To Fair Value (Including Dividends) | 7.09% | 26.09% | 18.05% |
(Source: DK Research Terminal, FactSet)
The S&P has a 15% consensus total return forecast through the end of 2025, a 6% annual consensus rate of return.
FAST Graphs, FactSet
FAST Graphs, FactSet
In the next 18 months, analysts think GOOG can deliver almost 4X the returns of the S&P.
In the next 5.5 years? GOOG might be able to triple, while the S&P is expected to deliver about 40% returns.
For anyone comfortable with its risk profile, GOOG is a potentially strong buy.
There are no risk-free companies, and no company is right for everyone. You have to be comfortable with the fundamental risk profile.
Morgan Stanley
If GOOG's BARD chatbot were to replace the existing search empire, it would cost the company $24 billion per year.
That's a 40% hit to the bottom line, created by GOOG doing too good a job with its amazing AI chatbot.
How do we quantify, monitor, and track such a complex risk profile? By doing what big institutions do.
DK uses S&P Global's global long-term risk-management ratings for our risk rating.
The DK risk rating is based on the global percentile of how a company's risk management compares to 8,000 S&P-rated companies covering 90% of the world's market cap.
S&P's risk management scores factor in things like:
Classification | S&P LT Risk-Management Global Percentile |
Risk-Management Interpretation |
Risk-Management Rating |
BTI, ILMN, SIEGY, SPGI, WM, CI, CSCO, WMB, SAP, CL | 100 | Exceptional (Top 80 companies in the world) | Very Low Risk |
Alphabet | 93 |
Exceptional |
Very Low Risk |
Strong ESG Stocks | 86 |
Very Good |
Very Low Risk |
Foreign Dividend Stocks | 77 |
Good, Bordering On Very Good |
Low Risk |
Ultra SWANs | 74 | Good | Low Risk |
Dividend Aristocrats | 67 | Above-Average (Bordering On Good) | Low Risk |
Low Volatility Stocks | 65 | Above-Average | Low Risk |
Master List average | 61 | Above-Average | Low Risk |
Dividend Kings | 60 | Above-Average | Low Risk |
Hyper-Growth stocks | 59 | Average, Bordering On Above-Average | Medium Risk |
Dividend Champions | 55 | Average | Medium Risk |
Monthly Dividend Stocks | 41 | Average | Medium Risk |
(Source: DK Research Terminal)
No risk-free companies exist, but GOOG is one of the lowest-risk ways to earn double-digit market-beating returns in the future, according to S&P.
When the facts change, I change my mind. What do you do, sir?" - John Maynard Keynes
There are no sacred cows. Wherever the fundamentals lead, we always follow. That's the essence of disciplined financial science, the math behind retiring rich and staying rich in retirement.
Dividend Kings Automated Investment Decision Tool
GOOG is one of the few tech titans whose blockbuster returns this year are sanctified by the righteousness of fundamentals.
Its valuation was so low in October of 2022 that even a 50% rally has left its 15% undervalued and capable of 55% gains over the next 18 months.
GOOG can gain more in the next 18 months than delivered in 2023's red-hot AI tech mania year.
GOOG is so undervalued and growing so quickly thanks to strong execution in the Cloud it could triple in the next 5.5 years, delivering 5X better returns than the S&P 500!
If you want to buy a red-hot momentum Ultra SWAN that still represents sanity in an insane world, GOOG is one of your best choices.
Receive free Big tech updates
We’ll send you a myFT Daily Digest email rounding up the latest Big tech news every morning.
When the US Department of Justice’s complaint against Google goes to trial next month, it will be the first case to delve into the business practices of a big tech company since the US took on Microsoft a quarter of a century ago. If Judge Amit Mehta comes down on the side of the US, the remedies he orders could have a direct bearing on Google’s core business, opening the door for rivals to take a bigger slice of the mobile search market.
You wouldn’t know it from the share price. Alphabet, Google’s parent, is up 47 per cent this year, part of a powerful rally that has seen Big Tech lead the entire stock market higher. Ever since the “techlash” that set in around 2017, when politicians and regulators around the world began to look at reining in the power of the biggest tech companies, investors have had a new risk to handicap. The current stock market mood suggests they see very little danger.
It is easy to see why. Despite the sound and fury emanating from Washington in accurate years, regulators have yet to score any big antitrust wins against the tech companies and Congress has failed to advance any important new legislation. And despite levying a series of fines against Google, Brussels has done little to change competitive dynamics in the markets it dominates.
The EU’s new Digital Markets Act may present a bigger risk. But in the absence of new laws in the US, regulators there have been forced to try to stretch existing ones. The courts, though, are wary of limiting business practices that confer immediate consumer benefits, such as lower prices. Tech companies are quick to warn that messing with their current way of doing business could threaten free internet services and low-price digital goods that have been popular with millions of consumers.
The US Federal Trade Commission‘s failure last month to convince a judge to block Microsoft’s $75bn acquisition of gaming company Activision Blizzard has again underlined the unwillingness of US courts to act without clear harm to consumers, however much the companies’ competitors complain.
The case against Google turns on a batch of deals the company struck with handset and browser makers to make its search engine the default on devices and handsets that run its Android software.
Though some parts of the complaint were thrown out by Mehta this month, the trial will focus on an issue where Google could be vulnerable. Section two of the Sherman Act imposes a broad ban on any “exclusionary conduct” used to monopolise a market. The US succeeded with a similar complaint against Microsoft, which used exclusive contracts to promote use of its Internet Explorer browser and defeat browser maker Netscape.
But even if Google’s search deals shut out competitors, the company would still prevail if it can show pro-competitive intent for its conduct. Google argues that paying to make its search engine the default that users see on their devices is no different from the way makers of breakfast cereals pay for prominent placement on supermarket shelves. It also warns that if its own promotional deals are cut off and the courts prevent a normal business practice, then it could lead to a worse experience for consumers, including higher phone prices.
The legal test comes just as another big tech company appears set to face a complaint over its core business. There have been reports for weeks that the FTC is moving closer to filing a long-awaited lawsuit against Amazon’s ecommerce operations. The agency is said to be targeting Amazon’s treatment of the third-party sellers who use its online marketplace to reach customers. Amazon has long faced protests that it effectively forces these sellers to pay for extra services, such as its Fulfilment by Amazon storage and delivery, to ensure their products are given prominent placement in its marketplace.
As with Google, this appears to attack an important part of the core business. Independent sellers like this account for nearly a quarter of Amazon’s revenue, making this an important part of the company’s business. Yet Wall Street seems untroubled and Amazon’s shares have rebounded 57 per cent this year.
One calculation investors seem to be making is that even if the companies lose, they may be able to resolve the cases by adjusting the terms of some of their contracts rather than being forced into fundamental changes to their operations. Amazon, for instance, has already agreed to concessions in the EU and UK over how it deals with third-party sellers, with little impact to its business.
The Google trial will shine a light on the sort of practices that Big Tech’s critics claim have complained about for years. But if Wall Street is right, it won’t put a dent in the companies’ most profitable operations.
richard.waters@ft.com
Google has pulled its AI Test Kitchen app from the Play Store and the App Store to focus solely on the web platform.
The company launched the AI Test Kitchen experience last year to let users interact with projects powered by different AI models such as LaMDA 2. The first set of experiments included the model breaking down a goal into different subsets and talking about dogs to check if the system sticks to the topic.
Google confirmed the move to 9to5Google, which first noted the apps being pulled, and said that AI Test Kitchen will focus on just the web experience as it is easy to push updates on just one platform.
Last November, Google announced “Season 2” of the AI Test Kitchen with new experiments. But they were never rolled out. Currently, the Test Kitchen hosts only a solitary text-to-music language model experiment called MusicLM, which was announced earlier this year at Google IO.
This move is not very surprising given Google has a habit of shutting down apps and experiments without prior notice. Plus, given the rise of large language models (LLMs) and generative AI-focused tools like OpenAI’s ChatGPT and Anthtropic’s Claude, the company might want to focus more on testing features for its consumer products.
In May, during Google IO, the company announced a new portal called Google Labs, which allows users to sign up for generative AI-based experiments. Notably, this page also lists the aforementioned MusicLM experiment.
It’s on-brand for Google to make things confusing by having multiple products for AI experiments. Now, we have an AI Test Kitchen page with one experiment. A Google Labs page shows different projects like Search Labs, the company’s AI-powered note-taking project NotebookLM, AI-focused Workspace features, along with the MusicLM project.
Google has pulled its AI Test Kitchen app from the Play Store and the App Store to focus solely on the web platform.
The company launched the AI Test Kitchen experience last year to let users interact with projects powered by different AI models such as LaMDA 2. The first set of experiments included the model breaking down a goal into different subsets and talking about dogs to check if the system sticks to the topic.
Google confirmed the move to 9to5Google, which first noted the apps being pulled, and said that AI Test Kitchen will focus on just the web experience as it is easy to push updates on just one platform.
Last November, Google announced "Season 2" of the AI Test Kitchen with new experiments. But they were never rolled out. Currently, the Test Kitchen hosts only a solitary text-to-music language model experiment called MusicLM, which was announced earlier this year at Google IO.
This move is not very surprising given Google has a habit of shutting down apps and experiments without prior notice. Plus, given the rise of large language models (LLMs) and generative AI-focused tools like OpenAI's ChatGPT and Anthtropic's Claude, the company might want to focus more on testing features for its consumer products.
In May, during Google IO, the company announced a new portal called Google Labs, which allows users to sign up for generative AI-based experiments. Notably, this page also lists the aforementioned MusicLM experiment.
It's on-brand for Google to make things confusing by having multiple products for AI experiments. Now, we have an AI Test Kitchen page with one experiment. A Google Labs page shows different projects like Search Labs, the company's AI-powered note-taking project NotebookLM, AI-focused Workspace features, along with the MusicLM project.
We use technologies like cookies to store and/or access device information. We do this to Excellerate browsing experience and to show personalized ads. Consenting to these technologies will allow us to process data such as browsing behavior or unique IDs on this site. Not consenting or withdrawing consent, may adversely affect certain features and functions.
The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
The technical storage or access that is used exclusively for statistical purposes. The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
Google has pulled its AI Test Kitchen app from the Play Store and the App Store to focus solely on the web platform.
The company launched the AI Test Kitchen experience last year to let users interact with projects powered by different AI models such as LaMDA 2. The first set of experiments included the model breaking down a goal into different subsets and talking about dogs to check if the system sticks to the topic.
Google confirmed the move to 9to5Google, which first noted the apps being pulled, and said that AI Test Kitchen will focus on just the web experience as it is easy to push updates on just one platform.
Last November, Google announced "Season 2" of the AI Test Kitchen with new experiments. But they were never rolled out. Currently, the Test Kitchen hosts only a solitary text-to-music language model experiment called MusicLM, which was announced earlier this year at Google IO.
This move is not very surprising given Google has a habit of shutting down apps and experiments without prior notice. Plus, given the rise of large language models (LLMs) and generative AI-focused tools like OpenAI's ChatGPT and Anthtropic's Claude, the company might want to focus more on testing features for its consumer products.
In May, during Google IO, the company announced a new portal called Google Labs, which allows users to sign up for generative AI-based experiments. Notably, this page also lists the aforementioned MusicLM experiment.
It's on-brand for Google to make things confusing by having multiple products for AI experiments. Now, we have an AI Test Kitchen page with one experiment. A Google Labs page shows different projects like Search Labs, the company's AI-powered note-taking project NotebookLM, AI-focused Workspace features, along with the MusicLM project.