Alphabet's Dual-Pronged Strategy: Securing AI Dominance

Alphabet expands its Pentagon AI partnership and launches Gemini Embedding 2, backed by strong Q4 2025 earnings and a massive $175B+ 2026 investment plan.

Alphabet Inc. is advancing on two strategic fronts this week, each move reinforcing its ambition to establish its Gemini AI as the leading platform. The company is simultaneously deepening its partnership with the U.S. Department of Defense and launching its most sophisticated AI model to date, Gemini Embedding 2. Both initiatives share a unified objective: cementing Gemini's position across commercial and government sectors.

Solid Financial Footing Enables Aggressive Moves

The company's ambitious push is underpinned by robust financial performance. For the fourth quarter of 2025, Alphabet reported revenue of $113.83 billion, an 18% year-over-year increase that comfortably exceeded Wall Street's expectations. Its Google Cloud segment saw particularly strong growth, with revenue surging 48% to $17.7 billion, achieving an annualized run rate of $70 billion.

This strength fuels a significant capital expenditure plan. For 2026, Alphabet has budgeted between $175 billion and $185 billion for investments—more than double the $91 billion spent the prior year and substantially above the approximately $120 billion anticipated by analysts. The market has reacted cautiously to this aggressive spending outlook, with the stock declining roughly 11% over the past four weeks. Shares closed yesterday at €265.15, a level that remains about 20% above its 200-day moving average.

Whether this investment drive will expand margins is a key question for upcoming quarterly reports. A strong indicator is Google Cloud's backlog, which stands at $240 billion—a 55% increase from the previous year—suggesting underlying demand may justify the substantial outlay.

Expanding the Pentagon Partnership

A significant part of Alphabet's strategy involves a substantial expansion of its Gemini models within the U.S. Department of Defense. Through the Pentagon's GenAI.mil portal, civilian and military personnel will now be able to build custom AI agents for unclassified tasks. The platform already offers eight pre-built agents that automate functions such as summarizing meeting notes, planning budgets, and reviewing policy alignment with national defense strategy.

Adoption metrics are notable: Since December, 1.2 million Pentagon staff have used the AI chatbot, submitting 40 million queries and uploading over four million documents. According to Undersecretary Emil Michael, discussions are underway to eventually deploy Gemini agents on classified networks as well.

This expansion offers Google a strategic opening. Its competitor, Anthropic, is currently entangled in a legal dispute with the Trump administration after being labeled a supply chain risk. This situation provides Google an opportunity to capture a larger share of the rapidly growing defense budget allocated for AI infrastructure.

The collaboration is not without internal controversy. Over 100 Google AI employees previously petitioned Chief Scientist Jeff Dean to prohibit the use of Gemini for surveillance of American citizens or for autonomous weapon systems without human oversight. This stance has gained wider support, with approximately 900 Google and 100 OpenAI workers now having signed an open letter endorsing the same demands.

Launching a Groundbreaking AI Model

Concurrently, the company has unveiled Gemini Embedding 2, marking its first natively multimodal embedding model. This technology processes text, images, video, audio, and documents within a single, unified embedding space. It is now accessible in public preview via the Gemini API and Vertex AI.

Supporting more than 100 languages, the model enhances applications like semantic search, sentiment analysis, and data clustering. For enterprise clients, it lowers the technical barrier to constructing complex data pipelines, which historically required separate processing steps for different media types. Initial benchmark tests indicate that Gemini Embedding 2 outperforms leading competitor models across text, image, and video categories.