/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

A look at the economics of Microsoft and Google inserting LLMs into search, crushing profitability and requiring massive capex, and the impact on the companies

SemiAnalysis :

SemiAnalysis

Context & Ripple Effects

Google had signaled that LaMDA would become a companion to search, putting the company and Microsoft on a path to make model inference part of a high-volume consumer product. SemiAnalysis frames the immediate constraint as the cost of doing so, rather than model availability alone.

Later coverage of nearly $80B in quarterly AI-infrastructure spending by Google, Meta, and Microsoft shows how the search-era compute burden became part of a broader capital-investment cycle.

First-order effects

  • Google and Microsoft face lower search profitability when LLM responses add recurring inference expense to each query, forcing both companies to absorb higher operating costs or limit how broadly they deploy the feature.
  • The two companies must commit substantial capital to the infrastructure supporting LLM-enabled search, shifting resources toward AI capacity.

Second-order effects

  • Google and Microsoft’s search competition becomes partly a contest over inference efficiency and available compute, because richer model features carry a larger cost burden at scale.
  • The need to fund search AI reinforces the infrastructure race already visible in record combined capex among major platforms, concentrating spending power among companies able to finance it.

Third-order effects

  • If LLM answers become a standard search interface, inference becomes a durable cost of goods sold for search rather than a temporary product-development expense.
  • Search economics may increasingly favor firms that can pair consumer distribution with large-scale infrastructure, making capital intensity a product constraint as well as a finance issue.

The trend: Search is evolving into an answer-engine market in which model-inference costs and infrastructure capacity shape product strategy as much as relevance does.

Discussion

  • Vox Kelsey Piper on x
    Are we racing toward AI catastrophe?
  • @hkanji Hussein Kanji on x
    If the ChatGPT model were ham-fisted into Google's existing search businesses, the impact would be devastating. There would be a $36 Billion reduction in operating income. This is $36 Billion of LLM inference costs. https://open.substack.com/...
  • @dylan522p Dylan Patel on x
    Imagine a dystopia where the rich suburban mom has access to the best search engine, and everyone else gets lower-cost ones... Given the cost per inference of LLMs, Google and Microsoft Bing have a decent argument for deploying them only to the highest CPM users... $GOOGL $MSFT
  • @dylan522p Dylan Patel on x
    More details on the actual cost structure. In reality, we probably just get paid search options that are much better than free. https://twitter.com/...
  • @01core_ben Ben Buchanan on x
    @dylan522p Semi Analysis is one of my absolute favorite newsletters. This most recent post on the Inference Cost of Search Disruption/LLM Cost analysis is amazing. Highly recommend reading and subscribing as well! https://www.semianalysis.com/ ...
  • @dylan522p Dylan Patel on x
    The Inference Cost Of Search Disruption $30B Of $GOOGL Profit Evaporating Overnight! ChatGPT and Google Search cost structure analysis Performance improvement with H100, TPUv4, TPUv5 Next-generation LLM architecture Latency optimization $MSFT $NVDA $AVGO https://www.semianalysis.…