I’m curious why and how ChatGPT kills search and revenue streams? Firstly, ChatGPT requires logging in. I’d rather use a service that doesn’t require logging in. Secondly, is the answer fast? Google once won the search engine race against Yahoo and Alta Vista because it had a blank search page and no ads, which made it clearly faster. I believe people still want to Google quickly. Thirdly, when Googling, often the first suggestion is not what one wants. It’s easier to browse through alternatives than to wait a minute for text that rambles and requires a new question.
4 Likes
This has certainly been touched upon in this thread as well, but I’ll bring my own perspective. Nowadays, if I’m looking for some information, I ask it directly from some AI, usually ChatGPT. So in this case, AI has completely replaced the use of Google. Anecdotally, I could also answer that it doesn’t matter to me if I have to log in to some service. You can give commands to AIs or agents, for example, to keep the answer short, in which case the answer is usually short 
Mainly, I’m trying to make the point that using AI is significantly more effortless than using Google when searching for information.
But I still use Google, but nowadays it’s merely a phone book that I use when I want to go to a specific site, for example, an online store.
7 Likes
Interesting discussion about this giant. I once wrote a brain dump on the slippery slopes after the investment cast’s Microsoft vs Google video.
I’ll try to write an updated brain dump vol. 2, tongue-in-cheek and with my cap backward.
- Let’s analyze where we stand in light of the Q3/2025 report and the Gemini 3.0 launch. An impossible task lies ahead.
- From an investor’s perspective, the situation is a bit exciting: the core business grinds out cash flow, but there’s always a slight clenching in the glutes.
1. Gemini 3.0 – Will Search Remain Relevant?
The discussion has shifted from chatbots to “agents.” Google has technically caught up with competitors and is at least on par, possibly even ahead, at least temporarily.
Agents: Gemini is already capable of performing valuable actions (reservations, purchases). This is indeed a very important development step, where Google is starting to fully leverage its current ecosystem and user base.
However, a critical question is how all of this will actually be commercialized profitably.
- Q3-data: “Google Search & Other” generated $56.56 billion in revenue with +14.5% growth. Search isn’t melting down; it’s growing. This is quite intense.
- The cost structure of generative search is certainly heavier, but it appears that ads embedded in answers are working. The company will surely have to balance user experience and margins.
- Quoting from Bloomberg’s news on download numbers:
”As of October, Gemini’s app had 73 million monthly downloads, well shy of ChatGPT’s 93 million monthly downloads, according to research firm Sensor Tower.”
In the bustling AI discussion, it’s important to note where users are actually performing financially valuable actions.
As a jest, with my cap askew:
I can write a poem for Fuge’s hottest travel guide with ChatGPT and create a greeting card at the same time.However, when I start putting money on the table and comparing hotels, flights, or Black Friday deals, I know which way to go. When researching winter tires, I can also ask Gemini for its thoughts. This activity might also bring joy to advertisers. I’m not saying a poem for Fuge’s travel guide isn’t valuable.
2. Infra: Own Chips (TPU) vs. Nvidia
-
TPU: The majority of the company’s own AI workload runs on Google’s TPU chips. This “protects” margins, as competitors pay for Nvidia’s hardware.
-
The flip side: TPU is partly a closed ecosystem. At least, that’s what I’ve understood? Feel free to correct me. The risk here is how development progresses and whether Google becomes isolated while the rest of the world develops applications with Nvidia’s standards.
-
Bloomberg’s news on Meta’s interest (same source as above):
”On Monday, tech publication the Information reported that Meta planned to use Google’s chips in its data centers in 2027. Google declined to address the specific plans, but said that its cloud business is “accelerating demand” for both its custom TPUs and Nvidia’s graphics processing units. “We are committed to supporting both, as we have for years,” a spokesperson wrote in a statement.” -
Here too, perhaps in the big picture, a positive factor can be found. The cost advantage is possibly on Google’s side. In a situation of fierce competition and profitability, this again brings an advantage to Google’s coffers. A short quote on this:
”Google’s TPUs are mostly attractive to a handful of companies with big computing bills, like Meta and Anthropic, said Meryem Arik, CEO of the AI startup Doubleword.”Business Insider article: Nvidia, AMD Stocks Fall As AI Chip War With Google Heats up - Business Insider
3. YouTube is currently, I believe, the cornerstone of advertising sales.
-
Q3/2025 YouTube advertising $10.26 billion (+15% growth). This is very impressive development. At the same time, it should be noted that more time is being spent on TikTok on mobile devices. It remains to be seen how Shorts develops and performs here.
- TV vs. Mobile: According to Nielsen data, YouTube wins (Connected TV) in the United States. https://www.nielsen.com/news-center/2025/youtube-netflix-ride-the-wave-of-summer-streaming-highs-in-nielsens-media-distributor-gauge/
-
Data: YouTube’s strategic importance is emphasized as a source of video-format training material for AI.
- This is certainly another positive resource as competition shifts to multimodal AI (video, audio, image).
4. New Growth Drivers: Cloud, Pixel, and (Waymo)
The company’s dependence on advertising has slightly decreased; new areas take time and still require investments.
-
Google Cloud: Q3/2025 figures ($15.15 billion (+33.5% growth)) confirm that the cloud is indeed the strongest growth platform. The competitive landscape is established: AWS leads in volume, Microsoft in enterprise integrations, Google seeks growth with an AI edge.
- ”Google Cloud is one of the most important priorities for Alphabet as a whole and I expect it to play an even more central role as the company moves forward,” Alphabet CEO Sundar Pichai told Reuters in an interview earlier in October.”
-
Pixel: Never heard of it? Q3/2025 “Google Subscriptions, Platforms and Devices” brought in $12.9 billion (+20.8%).
-
Pixel is also an important platform where Google can manage the development of hardware, software, and AI. Market share growth is positive, but the most important financial leverage is still search performed on its own device versus that done on an iPhone, for which a “tax” must be paid to Apple.
”Pixel phones do seem to be gaining steam. Sort of. Five years ago, only about 1 percent of American phone buyers picked a Pixel. Now it’s 4 percent.” https://www.washingtonpost.com/technology/2025/08/19/best-android-phone-google-pixel-apple-samsung/
-
In the “Luxury” class:
“Google’s share in the $600+ price band rose to 6.1% in September 2025 from 0.1% in September 2022, an impressive feat in the mature US market.”
Google Sees Record Single-Month Pixel Sales in US During September 2025
-
-
-
Waymo: Scale is growing (nearly 250k weekly rides), but global profitability and scalability remain a question mark. This is an option whose value can be anything between 0 and the sky. Currently, this is certainly not factored in when considering valuation. It’s utterly impossible to assign any direct value to it. Nevertheless, a nice bonus.
If you want some bullish reading, here it is: Waymo Is A Trillion-Dollar Opportunity. Google Just Needs To Seize It.
Let’s conclude with an estimate based on horoscopes and astronomy of how revenue might develop and what the structure might be, for example, three years from now in 2028:
-
Ad Business (67.99% of revenue):
Growth to approximately 5% level. Search would thus have matured significantly, and AI eats into volume, but price compensates. As a reminder, Q3 saw quite a strong performance with +14.5% growth. -
Cloud (20% of revenue):
This is a growth engine (approximately 20% growth rate). The current growth rate is (+33.5%). At the current pace, a 20% share of revenue could likely be achieved in 2027. -
Other (12.01% of revenue), i.e., Waymo/Pixel/Subs
The share of advertising in revenue will surely gradually drop from the current 75% to below 70%. The company will transform more into an infrastructure player.
Margins may possibly drop in the near future, but only time will tell. On the other hand, does the company have the potential to streamline its already efficient working methods and development processes?
An interesting situation could arise IF TPU sales really start taking off externally, towards Meta and partners. I’ll take my medicine in the meantime, though.
Many positive drivers are hidden in the text, but the intention was not to write a bullish analysis. Very strong holding. Wiser men are also in on it:
(Image: Voronoiapp and Author stockmen9)
The elephant eating is now over. Let’s just say that if I learned one thing, it’s that I won’t be writing about brain fart vol 3 for a while.
16 Likes
Waymo begins “manual” test drives for robotaxis in Baltimore, Pittsburgh, and St. Louis.
Waymo already operates, plans service, or tests its vehicles in at least 26 cities.
The new areas expand the service as Waymo aims for fully autonomous rides - expansion continues as competition intensifies.
https://www.cnbc.com/2025/12/03/waymo-baltimore-pittsburgh-stlouis.html
7 Likes
Google Cloud entered into a multi-year partnership agreement with AI coding platform Replit to strengthen its position in the rapidly growing vibe-coding market.
Replit will more extensively adopt Google’s cloud services and models, which can bring Google more new enterprise customers. This rapidly growing startup is expanding quickly and is already at a billion-dollar revenue level.
"
Points
- Google Cloud is locking in a multi-year partnership with AI coding startup Replit
- Google is betting on Replit as a breakout platform in the fast-growing vibe-coding phenomenon
- Replit will expand use of Google Cloud services, add Google models onto its platform, support AI coding use-cases"
https://www.cnbc.com/2025/12/04/google-replit-ai-vibe-coding-anthropic-cursor.html
5 Likes
10 Likes
U.S. judge Amit Mehta finalized the remedies concerning Google’s search monopoly and added new details to the decision on Friday.
The harshest penalties, such as the forced sale of Chrome, were rejected, which relieved investors. However, Google is not allowed to make long-term or exclusive agreements favoring its search engine, and must also open up more search data to competitors. According to the case, the decision doesn’t truly crush Google, but it can still “sting” nonetheless.
https://www.cnbc.com/2025/12/05/judge-finalize-remedies-in-google-antitrust-case.html
4 Likes
I would say there is no significant impact on Alphabet. The damage has already been done, meaning Google has gained the most popular position as a search engine and has accessed the most extensive data to train AI before others. The fact that it is no longer allowed to make competition-excluding agreements, e.g., with Apple, is not very relevant in my opinion. A legal loophole will surely be found for this as well.
2 Likes
Here’s another SalkunRakentaja article about Gemini’s progress and how ChatGPT’s progress has slowed down while still clearly being the “biggest.”
At the same time, Google’s Gemini is rapidly building its own user base. Alphabet CEO Sundar Pichai stated in the autumn earnings release that Gemini’s standalone application has reached over six hundred and fifty million monthly users, whereas in the spring, the number was only around three hundred million. The growth has been exceptionally rapid, driven not only by aggressive distribution within Google’s ecosystem but also by image generator features that have attracted consumers, such as the Nano Banana function.
Subheadings:
- User growth slows – engagement diversifies
- Gemini leverages its distribution advantage
- Heavy cost structure weighs on OpenAI
- Google plays the long game with results
- Competition shifts from consumers to infrastructure
7 Likes
Alphabet is specifically allowed to make exclusive agreements as long as their duration is a maximum of one year. Previously, the duration of these agreements was not limited. This entire anti-trust lawsuit has turned out favorably for Alphabet. I would characterize these more specific terms as very lenient, and previously, there was no need to divest the Chrome browser either. In my opinion, these exclusive search agreements are specifically very important for the company’s customer acquisition, especially for low-cost acquisition. Now, instead of a large 10 billion price tag, an annually varying amount of money is paid for exclusivity. This practically enables a similar operating model compared to previous years, although the price tag for exclusivity receives inflationary pressure annually during a rising market.
Data sharing also apparently remains a small part, and surely our beloved monopoly will throw a wrench in the works for anyone who wants to get their hands on this raw data. If a better scenario existed regarding this lawsuit, I certainly didn’t imagine it myself. In practice, this was the judge’s approval of the monopoly. From this, it is good to continue protecting business operations.
7 Likes
Google is bringing developers a new Antigravity environment, which is built upon the Gemini 3 Pro model, and the article states that the company is putting AI agent coding at the forefront.
High demand forced adjustments to usage limits; for example, Pro and Ultra subscribers receive the highest quotas, while for free users, Antigravity is distributed as weekly quotas, which are consumed according to the task’s complexity, rather than the number of requests.
The entire Gemini 3 suite is experiencing high user demand, with Google capping Gemini 3 Pro and Nano Banana Pro usage at the end of November. Now, Google is making similar adjustments for Antigravity. However, instead of slashing rate limits, the company is tweaking them to make Antigravity available as much as possible while keeping up with the demand.
5 Likes
This tweet below might be of interest in this thread, although far-reaching conclusions cannot, of course, be drawn based on ChatGPT’s steeper decline compared to others. 
https://x.com/Similarweb/status/1997963677065195902
On the other hand, it was doing great in September and October
3 Likes
Personally, my use of ChatGPT’s website has significantly decreased, as I use the app on my phone and Copilot’s GPT5 model on my computer.
Just as a note, these website visits might not be as relevant in the future.
7 Likes
Good point, initially many only used it on a computer, but now many are taking out their phones with their apps.
Below is an article about how Google plans to bring its first AI glasses to market next year to compete with Meta’s Ray-Ban Meta glasses.
Coming are both voice-assisted Gemini glasses and also models with a display embedded in the lens for map and translation information, for example. The devices are based on the Android XR model and are being developed in collaboration with Warby Parker, Samsung, and Gentle Monster, among others.
https://www.cnbc.com/2025/12/08/google-ai-glasses-launch-2026.html
4 Likes
The article below discusses a familiar topic, detailing, for example, how Google’s Gemini is rapidly catching up to OpenAI’s ChatGPT, doubling web visits (August–November 2025) and increasing monthly active users to 346 million according to recent data.
ChatGPT still maintains its lead with approximately 810 million users and generally longer user sessions. The competition is intensifying to such an extent that OpenAI’s Sam Altman has already declared an internal “code red” within the company.
In addition, Google plans to strengthen its position with the search engine’s AI functionalities and its rapidly growing Gemini application, which is already widely used around the world, etc.
9 Likes
This probably isn’t about large sums of money in terms of dollars, at least not yet, but it’s still likely interesting news.
The U.S. Department of Defense is adopting Google’s Gemini AI model as the first on the Pentagon’s new GenAI.mil platform, which is intended for three million employees.
The model will be used for background research, document formatting, and also image analysis, among other things. However, Google’s previous $200 million AI contract with the Department of Defense is already in the background.
U.S. Defense Secretary Pete Hegseth said on Tuesday that the Department of Defense had chosen Alphabet Inc’s (NASDAQ:GOOGL) Gemini as the first model in an artificial intelligence platform for its three million employees.
9 Likes
News about regulatory threats are probably tiring here, but here’s a pretty good critical article about Alphabet’s opportunity to “exploit” the entire internet for AI training for free with Googlebot, while others pay for it. In addition, when the AI answer fills the top of Google, virtually all pages are left in the dark.
In the old days, it was joked that the best place to hide a body is to put it on the second page of Google search results. Now it doesn’t matter if the body is in the place of the first link, because the AI answer covers it.
5 Likes
I predict that at least one thing will happen eventually:
A way for websites to allow Google search indexing but prevent AI training bots from snooping.
Currently, these are bundled, and websites have no way to block only AI training from Google. And blocking search indexing is practically a guarantee that no one will ever visit your site.
For other actors, AI crawlers can be blocked, at least in theory.
On the other hand, this is the fundamental problem of AI services in a nutshell. They only work if they can train freely on everything they can get their hands on. But allowing this for content-producing entities destroys their business. A solution must be found for this eventually, because otherwise either AI bots will be blocked from all content, or all content production will cease because it is no longer economically viable, when the only reader is an AI training bot and people read the material from an AI service.
9 Likes
I have been following the development of robotaxis and Waymo’s scaling with interest. Waymo has been predicted to be a loser in robotaxis because it cannot scale like Tesla. Currently, however, it seems that Waymo’s expansion is succeeding well.
https://www.cnbc.com/amp/2025/12/08/waymo-paid-rides-robotaxi-tesla.html
In a year, weekly rides have grown from 150k rides to 450k rides. Costs are certainly also increasing here, because the cars are more expensive than Tesla’s counterparts, but I still believe that Waymo can push costs significantly lower as operations expand. Waymo also has the first-mover advantage, meaning it can capture markets from others and establish its brand in consumers’ minds. Interesting!
11 Likes
Google DeepMind plans to open its first “automated research laboratory” in the UK next year.
The laboratory will combine artificial intelligence and robotics for the development of new superconductors and semiconductor materials.
British researchers will gain privileged access to DeepMind’s advanced AI tools, and later, cooperation could potentially expand to, for example, nuclear fusion research. The project is part of the UK’s strategy to strengthen the country’s AI expertise.
https://www.cnbc.com/2025/12/11/googles-ai-unit-deepmind-announces-uk-automated-research-lab.html
4 Likes