● Apple-Google Shock Alliance, Meta Mega Buildout, DeepSeek Sparks DDR5-CXL Memory Boom
This article contains three key takeaways that will overturn the economy and AI trends in the first half of 2026, going beyond simple news delivery.
In particular, we analyzed the changes in the mobile market brought about by the massive alliance between Apple and Google, Meta’s unimaginable infrastructure investment, and most importantly, the opportunity for new memory semiconductors pointed out by the DeepSeek paper.
When others simply say “AI has developed,” we specifically point out which stocks and industries this change will cause money to flow into.
Check it out right now.
1. The Apple-Google Alliance: The Hegemony of Mobile AI Changes
The Secret Behind Alphabet (Google) Surpassing a $4 Trillion Market Cap
News is shaking the market as Apple and Google (Alphabet) join hands.
This is because Google’s Gemini model was selected as the core partner for Apple’s next-generation ‘Apple Intelligence’.
With just this news, Alphabet’s market cap soared vertically, leaping once again into a giant enterprise approaching about 5,000 trillion won.
Why is this important? Looking at AI traffic share changes over the past year gives the answer.
In a market where OpenAI (ChatGPT) was dominating, Google’s Gemini has risen at a frightening speed and is now competing on equal footing.
Google possesses massive financial power and its own cloud, so it can be seen as having occupied an advantageous position in the long game compared to OpenAI, which needs external funding.
The Real Reason Apple Chose Gemini Instead of ChatGPT
Why did Apple choose Google over OpenAI?
According to Apple’s joint statement, they determined that “after careful evaluation, Google’s AI technology provides the most capable foundation.”
This seems to be a result considering not only technological prowess but also stable infrastructure and long-term partnerships within the global AI trend.
You know the reason we can’t leave the iPhone is the ‘Data Lock-in’ effect, right?
Now, Siri combined with Gemini will completely learn your emails, schedules, and text history to be reborn as a personalized assistant.
It reaches a level where if you ask, “Where was that dinner restaurant I promised my friend last week?”, it gives an answer immediately.
If this is realized, stagnant smartphone replacement demand will explode, and the on-device AI market will heat up once again.
2. Meta’s Declaration of War: The Ultimate King of Infrastructure Investment
Construction of Data Centers on a Scale of Tens of Gigawatts (GW)
Mark Zuckerberg recently announced a new initiative called ‘Meta Compute’.
The core point is to expand data centers to 5-6 times the current size over the next 10 years, and over 20 times in the long term.
It is estimated that Meta currently holds about 5-7GW, and they mean to increase this to the 20-50GW level.
This isn’t simple expansion, but an expression of the will not to be pushed back in the AI hegemony competition.
Data center infrastructure expansion inevitably leads to power facilities, cooling systems, and massive hardware purchases.
The fact that big tech companies are pouring money in like this is strong evidence that the growth of the AI industry is only at the starting stage.
3. [In-depth Analysis] The ‘Real’ Opportunity the DeepSeek Paper Speaks Of
It’s Not Simple Efficiency: The Landscape of Memory Semiconductors Changes
This part is the core point of today’s article and content not well covered in other news.
China’s DeepSeek published a new paper, and the market worried like last year, “Isn’t AI becoming too efficient, reducing semiconductor demand?”
However, the conclusion of this paper is quite the opposite.
Rather, it is a tremendous boon for the memory semiconductor market, especially DDR5 and CXL-related stocks.
The key takeaway of DeepSeek’s paper is ‘Conditional Memory’ technology.
I’ll explain it easily.
AI Functioning Like a Human Brain: Role Division Between HBM and DDR5
Previously, trying to cram all knowledge into expensive and capacity-limited HBM (High Bandwidth Memory) meant there was a limit to increasing model size.
But DeepSeek changed the brain structure.
- Thinking Brain (HBM): Handles only logical thinking or core calculations that need immediate processing.
- Giant Library (CXL/DDR5): Detailed knowledge (e.g., the year of the Imjin War, specific historical facts, etc.) is stored in general DRAM, which has large capacity and is cheap.
When AI receives a question, it usually uses only HBM, but goes to the library (DDR5) to take out a book only when detailed knowledge is needed.
If this happens, the problem of HBM capacity shortage is solved, and the size of the AI model can be increased virtually infinitely.
Ultimately, the conclusion is reached that “HBM is still insufficient and needed a lot, but now general server DRAM (DDR5) is also needed in incredibly large quantities.”
This flow matches exactly with what Nvidia CEO Jensen Huang recently said: “Now memory is the bottleneck, and it is the era of memory.”
Now, market interest is very likely to expand beyond simple GPUs to optical communication and high-capacity memory interfaces (CXL).
< Summary >
- Apple-Google Alliance: As Apple selects Google Gemini as its next-gen AI partner, Alphabet’s market cap skyrockets, forecasting a reorganization of the mobile AI market.
- Meta’s Infrastructure Investment: Declaring to expand data center scale by tens of times over the next 10 years, investment in AI hardware and power infrastructure will continue.
- Reinterpretation of DeepSeek Paper: Separating the structure of AI models into ‘Thinking (HBM)’ and ‘Knowledge Storage (DDR5/CXL)’. This means an explosive increase in demand for DDR5 and next-gen memory, not a decrease in semiconductor demand.
- Investment Perspective: HBM is basic, and now is the time to pay attention to the value chain related to DDR5 and CXL.
[Related Articles…]
- Apple and Google’s AI Alliance, Analysis of the Impact on the Market
- DeepSeek Paper Explanation: Why are Memory Semiconductors Hot Again?
*Source: 내일은 투자왕 – 김단테
● Apple-Google Shock Alliance, Meta Mega Buildout, DeepSeek Sparks DDR5-CXL Memory Boom
This article contains three key takeaways that will overturn the economy and AI trends in the first half of 2026, going beyond simple news delivery.
In particular, we analyzed the changes in the mobile market brought about by the massive alliance between Apple and Google, Meta’s unimaginable infrastructure investment, and most importantly, the opportunity for new memory semiconductors pointed out by the DeepSeek paper.
When others simply say “AI has developed,” we specifically point out which stocks and industries this change will cause money to flow into.
Check it out right now.
1. The Apple-Google Alliance: The Hegemony of Mobile AI Changes
The Secret Behind Alphabet (Google) Surpassing a $4 Trillion Market Cap
News is shaking the market as Apple and Google (Alphabet) join hands.
This is because Google’s Gemini model was selected as the core partner for Apple’s next-generation ‘Apple Intelligence’.
With just this news, Alphabet’s market cap soared vertically, leaping once again into a giant enterprise approaching about 5,000 trillion won.
Why is this important? Looking at AI traffic share changes over the past year gives the answer.
In a market where OpenAI (ChatGPT) was dominating, Google’s Gemini has risen at a frightening speed and is now competing on equal footing.
Google possesses massive financial power and its own cloud, so it can be seen as having occupied an advantageous position in the long game compared to OpenAI, which needs external funding.
The Real Reason Apple Chose Gemini Instead of ChatGPT
Why did Apple choose Google over OpenAI?
According to Apple’s joint statement, they determined that “after careful evaluation, Google’s AI technology provides the most capable foundation.”
This seems to be a result considering not only technological prowess but also stable infrastructure and long-term partnerships within the global AI trend.
You know the reason we can’t leave the iPhone is the ‘Data Lock-in’ effect, right?
Now, Siri combined with Gemini will completely learn your emails, schedules, and text history to be reborn as a personalized assistant.
It reaches a level where if you ask, “Where was that dinner restaurant I promised my friend last week?”, it gives an answer immediately.
If this is realized, stagnant smartphone replacement demand will explode, and the on-device AI market will heat up once again.
2. Meta’s Declaration of War: The Ultimate King of Infrastructure Investment
Construction of Data Centers on a Scale of Tens of Gigawatts (GW)
Mark Zuckerberg recently announced a new initiative called ‘Meta Compute’.
The core point is to expand data centers to 5-6 times the current size over the next 10 years, and over 20 times in the long term.
It is estimated that Meta currently holds about 5-7GW, and they mean to increase this to the 20-50GW level.
This isn’t simple expansion, but an expression of the will not to be pushed back in the AI hegemony competition.
Data center infrastructure expansion inevitably leads to power facilities, cooling systems, and massive hardware purchases.
The fact that big tech companies are pouring money in like this is strong evidence that the growth of the AI industry is only at the starting stage.
3. [In-depth Analysis] The ‘Real’ Opportunity the DeepSeek Paper Speaks Of
It’s Not Simple Efficiency: The Landscape of Memory Semiconductors Changes
This part is the core point of today’s article and content not well covered in other news.
China’s DeepSeek published a new paper, and the market worried like last year, “Isn’t AI becoming too efficient, reducing semiconductor demand?”
However, the conclusion of this paper is quite the opposite.
Rather, it is a tremendous boon for the memory semiconductor market, especially DDR5 and CXL-related stocks.
The key takeaway of DeepSeek’s paper is ‘Conditional Memory’ technology.
I’ll explain it easily.
AI Functioning Like a Human Brain: Role Division Between HBM and DDR5
Previously, trying to cram all knowledge into expensive and capacity-limited HBM (High Bandwidth Memory) meant there was a limit to increasing model size.
But DeepSeek changed the brain structure.
- Thinking Brain (HBM): Handles only logical thinking or core calculations that need immediate processing.
- Giant Library (CXL/DDR5): Detailed knowledge (e.g., the year of the Imjin War, specific historical facts, etc.) is stored in general DRAM, which has large capacity and is cheap.
When AI receives a question, it usually uses only HBM, but goes to the library (DDR5) to take out a book only when detailed knowledge is needed.
If this happens, the problem of HBM capacity shortage is solved, and the size of the AI model can be increased virtually infinitely.
Ultimately, the conclusion is reached that “HBM is still insufficient and needed a lot, but now general server DRAM (DDR5) is also needed in incredibly large quantities.”
This flow matches exactly with what Nvidia CEO Jensen Huang recently said: “Now memory is the bottleneck, and it is the era of memory.”
Now, market interest is very likely to expand beyond simple GPUs to optical communication and high-capacity memory interfaces (CXL).
< Summary >
- Apple-Google Alliance: As Apple selects Google Gemini as its next-gen AI partner, Alphabet’s market cap skyrockets, forecasting a reorganization of the mobile AI market.
- Meta’s Infrastructure Investment: Declaring to expand data center scale by tens of times over the next 10 years, investment in AI hardware and power infrastructure will continue.
- Reinterpretation of DeepSeek Paper: Separating the structure of AI models into ‘Thinking (HBM)’ and ‘Knowledge Storage (DDR5/CXL)’. This means an explosive increase in demand for DDR5 and next-gen memory, not a decrease in semiconductor demand.
- Investment Perspective: HBM is basic, and now is the time to pay attention to the value chain related to DDR5 and CXL.
[Related Articles…]
- Apple and Google’s AI Alliance, Analysis of the Impact on the Market
- DeepSeek Paper Explanation: Why are Memory Semiconductors Hot Again?
*Source: 내일은 투자왕 – 김단테
