● Buffett Final Gambit – Google Crushes Nvidia in AI Cost War

Buffett’s Final Gambit: The Real Reason Behind Buying Google (Alphabet) and the Winner of the AI Cost War

Today’s article covers all aspects, including Buffett’s large-scale acquisition of Google (Alphabet) revealed through the 13F filing, an analysis of Google’s AI industry cost structure (“TPU vs. NVIDIA”), an investigation into the controversy over inflated search queries, and strategies for valuation, risk, and portfolio management.
In particular, the “long-term cost advantage created by vertical integration of AI infrastructure” and the “checklist for verifying search traffic quality” are key points that are not well covered in other news outlets or YouTube.
It also connects how the risk-reward dynamics in the stock market are changing amid global economic uncertainty, debates over peak interest rates, and periods of slowing inflation.

[Breaking Summary] What Berkshire’s 13F Revealed

According to Berkshire Hathaway’s recent quarterly 13F filing, Alphabet appears to have been newly and significantly added to the portfolio.
While the purchase might have been executed by Ted/Todd managers, the inclusion of a top-tier portfolio position strongly suggests alignment with Warren Buffett’s philosophy.
Market consensus reads it as “Buffett’s final big move.”
Similar to the case with Apple, there is also a possibility of initially a small stake followed by gradually increasing the position.

Why Google Now: Buffett’s Understanding of Media and Cash Generation

Although Buffett has been cautious with tech stocks, he is well-versed in the economics of the media industry.
Search and YouTube are key media assets in the digital age, characterized by high barriers to entry and powerful network effects.
The robust free cash flow generated from advertising, cloud, and platform businesses provides defensive strength against cycles in the economy and interest rates.
Even amidst debates over global economic slowdown and re-ignited inflation, the structural cash generation aligns with the core criteria of Buffett-style value investing.

Google’s Current State: Overcoming “AI Fears” and Regaining Performance Momentum

Although there were significant concerns about search erosion immediately following the emergence of ChatGPT, recent quarterly revenue and advertising revenue have rebounded with consistent double-digit growth.
Search remains outstanding in advertising conversion efficiency, thanks to its intrinsic strength in capturing user intent.
YouTube is absorbing demand from retail, app installations, and branding, as short-form and CTV advertising grow simultaneously.
Google Cloud is simultaneously achieving revenue growth and improved profitability, buoyed by the demand for AI workloads.
In conclusion, even as AI competition intensifies, the core business and growth arm are operating in tandem.

The Essence of the AI Cost War: The Cost Advantage Created by TPU Vertical Integration

Google has built a vertical integration from research to models, frameworks, AI accelerators (TPUs), data centers, and services.
This strategically reduces dependence on NVIDIA and is highly effective in compressing per-token costs over the long term.
According to industry sources, even previous generations of TPUs are being reused at high utilization rates, thereby optimizing depreciation burdens.
There has been a consistent trend of increasing TPU production share at TSMC’s state-of-the-art packaging/nodes, establishing it as a secondary pillar in terms of hardware.
In terms of token processing capacity, batch efficiency, and total cost of ownership (including power and cooling), Google is securing a tangible advantage in large-scale inference.
This cost structure is likely to become increasingly differentiated as the AI cycle progresses into its later stages.

Product Updates: Gemini, Image, and Video Generation and Monetization

On public evaluation platforms and rankings, Google’s latest models maintain top-tier quality.
The improved quality of image and video generation models is driving automation in creator and marketing workflows.
User feedback on search “AI responses (Overview/Mode)” is improving, and gradual ad integration along with matching commercial intent queries are being attempted.
YouTube is redefining ad inventory and performance measurement through a combination of generative tools and advertising solutions.
Google Cloud is lowering the barrier to enterprise AI adoption with offerings such as Vertex, model accelerators, and vector databases.

Valuation Check: ‘Not Cheap’ vs. ‘Structural Improvement Premium’

By traditional forward multiples, it is valued near the historical upper range, and there is also an opinion that free cash flow-based multiples pose short-term burdens.
However, when factors such as an improved ad mix, cloud profitability, and margin leverage from AI cost reductions are considered, there is a persuasive argument for partly accepting a structural improvement premium.
If the interest rate regime stabilizes downward, lower discount rates on growth cash flows would favor multiple defenses.
Conversely, a resurgence in inflation and a rebound in long-term interest rates could act as risks that compress multiples.

Key Risk Check: Data, Regulation, Cycle

It’s about the controversy over inflated search queries.
Some researchers suggest that the activities of AI agents and bots may have distorted Google’s query and click metrics.
There are three test points:

• Whether there is a discrepancy between advertisers’ conversion metrics (ROAS, incremental lift) and platform-reported numbers.
• Whether the platform-reported metrics are abnormally rising compared to human-based measurements (panel, MMM).
• Whether there is a breakdown in the correlation between organic and paid traffic across brand and retail categories.
  It’s about regulatory and antitrust issues.
  There remains the possibility of regional sanctions in areas such as search distribution agreements, default apps on app stores and browsers, and data usage.
  It’s about the AI bubble and Capex cycle risks.
  If AI infrastructure investments become overheated, risks related to depreciation, power costs, and yield could increase performance volatility.
  Renegotiations of default search agreements with companies like Apple and Samsung are also variables.
  Intensified cloud competition may result in increased pressure to reduce prices.

Key Watch Points by Scenario: What Moves the Multiples

Under the bullish scenario:

• Ads naturally integrate into search AI responses, leading to simultaneous improvements in CPC and conversion rates.
• YouTube CTV and Shorts experience steady growth, and commerce integration is enhanced.
• Google Cloud’s margin levels are reassessed due to AI inference workloads.
• With TPU expansion, per-token costs decrease, fundamentally boosting AI service margins.
  For the base scenario:
• Search and YouTube achieve modest double-digit growth, Cloud maintains stable profitability, and AI Capex is managed at a reasonable level relative to revenue.
• Multiples remain in a trading range near historical medians.
  In a bearish scenario:
• The controversy over query quality leads to a decline in advertising efficiency, compounded by regulatory and contractual uncertainties.
• Increasing AI Capex burdens depress free cash flow valuations.

Portfolio Strategy: A Practical Checklist

Entry and position sizing strategy:

• If considering long-term holding, a staggered approach around earnings/guidance update dates is advisable.
• Increase the position when advertising, cloud, and AI cost metrics all improve simultaneously; if any of them weaken, adopt a wait-and-see approach.
  Monitoring indicators include:
• Advertiser conversion lift, YouTube CTV/Shorts RPM, cloud operating profit margin, Capex-to-revenue ratio, per-token cost estimates, and power usage effectiveness (PUE).
  Risk hedge strategies:
• In overheating AI phases, defend against volatility with low-beta value stocks and cash allocation, and consider pair trading with overheated AI hardware stocks if necessary.

A Compilation of Key Points Rarely Covered Elsewhere

• It’s a battle over unit costs in AI inference.
  Decreases in TPU-based inference TCO simultaneously boost margins across the “search AI response + YouTube generative + cloud hosting” triangle.
• The economics of depreciation.
  High reuse rates of previous generation TPUs mitigate the “hardware equals depreciation bomb” narrative and offer defensive strength during cyclical downturns.
• A paradigm shift in measurement.
  In AI response interfaces, “action-based conversion” is becoming more important than “clicks,” and the ability to conduct MMM and experimental design determines advertiser performance.
  This could serve as a catalyst for reevaluating Google’s adtech stack competitiveness.
• The network effect of data and computing.
  When large-scale first-party data is combined with low-cost computing, model update cycles accelerate and the product-revenue learning loop is enhanced.

Market Context: Interest Rates, Inflation, and Multiples

In a phase of peaking interest rates, the elevated present value of long-term growth cash flows supports higher multiples for platform stocks.
Conversely, if re-ignited inflation causes real interest rates to rise, it could lead to multiple compression.
During periods of heightened global economic volatility, the defensive strength is determined by the combination of cash generation and cost advantage.
Google, as one of the rare large-cap companies possessing this combination, aligns with Buffett’s selection criteria.

< Summary >

• Berkshire’s large-scale purchase of Alphabet is interpreted as a bet on cash generation and an AI cost advantage.
• Google is growing simultaneously in search, YouTube, and cloud, while its TPU vertical integration structurally reduces per-token costs.
• The controversies over query and click quality, regulatory and contractual issues, and AI Capex burdens constitute key risks.
• By managing risks with key monitoring indicators and a phased approach, the core of multiple defense lies in the trend of improved costs and margins.
• Multiples’ resilience can vary based on the path of interest rates and inflation, so macro factors must also be monitored.

[Related Articles…]
Google and TPU: Reading the Winner of the AI Cost War
The Future of Google Advertising: AI Overview Monetization Scenario

*Source: [ 내일은 투자왕 – 김단테 ]

– 버핏의 마지막 승부수 심층 분석

---