Meta pivots from Metaverse to AI glasses: The 'next platform' plan for 20 million units per year
The keyword defining tech news since the start of 2026 is the 'implementation phase' of AI. Moving away from 'future narratives' like the metaverse toward areas that sell now, are used now, and where social costs are becoming an issue—starting with discussions from Bloomberg Tech, we will organize the landscape from an investor's perspective, covering Meta's shift to AI devices, Microsoft's measures for regional 'burden' from data centers, the Trump administration's proposal for a credit card interest rate cap, and the geopolitics surrounding Taiwan and TSMC.
1. Meta: From Metaverse to 'AI Devices,' the winning strategy is 'daily flow'
In the Bloomberg program, it was discussed how Meta is scaling back metaverse-related projects (part of Reality Labs) and shifting its focus to wearables like Ray-Ban Meta glasses and data centers.Symbolic of this is the investor comment that 'it would be better for the company to remove "metaverse" from the headlines.'
Reports also suggest that Meta and EssilorLuxottica are planning to increase production capacity for Ray-Ban smart glasses to 20 million units per year, with the potential to reach over 30 million if demand continues.
On the other hand, Reality Labs has accumulated over $60 billion in losses since 2020, which is an important factor that created pressure for 'reallocation rather than withdrawal.'
1-1. Why glasses: Eliminating the 'hassle of launching' AI
With smartphone AI, the 'ritual' of opening an app, inputting data, and so on still remains. In contrast, as mentioned in the program, glasses are well-suited for 'voice-based AI,' allowing AI to be integrated into daily activities. This 'shortening of the flow' is directly linked to the potential for expansion into advertising, search, and commerce (Meta's core business).
2. Microsoft: Data centers are about what happens 'after they are built'—5 social contracts
Behind the AI infrastructure race, residents are facing issues with electricity costs, water, taxes, and employment. Microsoftarticulated '5 promises' in January 2026 as a Community-First AI Infrastructure, anticipating backlash from data center expansion.
The company's pillars are as follows:
Do not drive up electricity costs (the company will bear the necessary costs)
Minimize water usage and return more than the amount used to the region
Create local employment
Expand the tax base that benefits hospitals, schools, etc.
Strengthen the region through AI training and non-profit support
2-1. What investors should look at is 'permitting risk' rather than 'CAPEX'
Microsoft itself, while acknowledging the reality that AI requires massive amounts of electricity, cites IEA estimates that U.S. data center power demand will more than triple by 2035 (from 200 to 640 TWh/year).
In other words, the bottleneck for competition is not just GPUs, but power transmission, substations, water, and regional consensus. If friction occurs here, construction delays could lead to delays in revenue recognition. Microsoft's proactive move to take on 'local bills' can be interpreted as buying the prerequisite for growth (a social license).
3. Trump's '10% credit card interest cap': Who is the winner, and what are the points of contention?
In the program, President Trump touched on a proposal to cap credit card interest rates at 10% for one year, and the Klarna CEO supported it with strong words, saying 'Americans are being exploited.' As a policy, there is strong opposition from the financial industry, and the JPMorgan CFO has also warned that it could have 'negative consequences for consumers.'
The important point is that the discussion involves not only 'household relief' but also the risk of a contraction in credit provision (tightening of credit). Even now, the average credit card interest rate is approximately 20.97% (according to FRB data), and a 10% rate would shake the business model.
On the other hand, for Buy Now, Pay Later (BNPL) services, this could highlight their 'comparative advantage,' and we can foresee a scenario where operators like Klarna step to the political forefront.
4. Taiwan/TSMC: AI supremacy also resides in the 'number of factories' and 'tariff rates'
In the latter half of the program, it was reported that the U.S. is advancing trade talks with Taiwan, with a move to lower tariffs on Taiwan from 20% to 15%.
At the same time, TSMC's investment within the U.S. (additional construction in Arizona) is being discussed as a bargaining chip, and the Taiwanese side has also mentioned a 'broad consensus'.
The essence here is that the AI race has become inseparable from the political risks of the supply chain, not just 'model performance' or the 'number of data centers.' For investors, semiconductor-related news is now in a phase where price movements are more easily triggered by tariffs, subsidies, and relations with China than by 'technology' itself.
5. Summary: The main battleground of 2026 is 'AI x Hardware x Infrastructure x Politics'
If we connect this series of topics with a single thread, it looks like this.
Meta: Integrating AI into a 'daily use' format to capture the user journey (wearables)
Microsoft: Using costs to suppress 'regional backlash' that could halt AI growth (social contract)
Politics: Interest rate regulations and tariff negotiations could directly hit the revenue structures of payments and semiconductors
The 'AI market' can no longer be read solely through cloud earnings. By tracking these three elements—device penetration, power/water/permits, and policy—the resolution of the news will increase by one level.
