Hamidun News AI Agenda Index: First Half of 2026

Over 188 days — from January 12 to July 19, 2026 — Hamidun News processed 7734 materials on artificial intelligence from 91 sources. This is the first edition of our AI Agenda Index. Once per period, we show what the global and Russian AI press actually covered — according to our feed, without press releases or marketing.

The main finding of the half-year is unexpected. The most frequent topic in headlines is not a company, but a format. AI agents are mentioned in 892 headlines. This is more than OpenAI (768) and Anthropic (740) individually. The agenda has shifted from "whose model is smarter" to "what these models do on their own."

Who Sets the Agenda

We counted how often major players appear in headlines over the half-year — searching by company name in the headline across the entire array of 7734 materials:

• OpenAI — 768 headlines (9.9% of the entire feed)
• Anthropic / Claude — 740 (9.6%)
• Google / Gemini — 548 (7.1%)
• NVIDIA — 338
• Amazon / AWS — 271
• Meta — 212
• Microsoft — 162
• xAI / Grok — 147
• Apple — 139
• DeepSeek — 70

The gap between the first pair and the rest is nearly twofold. Two laboratories, OpenAI and Anthropic, hold one-fifth of all AI headlines. NVIDIA is the only "hardware" player in the upper part of the list: infrastructure remains a story, not just background.

Agents Surpassed Vendors

892 headlines with the word "agent" is not about one product, but about a shift in topic. A year ago, the agenda was driven by model releases. In the first half of 2026, the main story became what AI does without humans: autonomous assistants, code agents, task orchestration. The topic has surpassed any individual company.

What Was Written About in July

July yielded 1414 materials. By topics:

• Language Models (LLM) — 276
• Hardware — 254
• Business — 197
• Creative — 181
• Autonomous Systems — 156
• Startups — 90
• Medicine — 77
• Robotics — 70
• AI Ethics — 58
• Research — 55

Language models and "hardware" are nearly equal — behind each new generation of models is a race for chips and data centers. July's weekly pace grew: 343 materials in week 26, 409 in week 27, 662 in week 28.

Who We Read

Over the half-year, materials from 91 sources appeared in the feed. July's top:

• The Next Web — 248 materials
• Bloomberg Technology — 220
• Habr, AI section — 197
• 3DNews — 92
• TechCrunch — 58
• MarkTechPost — 51
• AWS Machine Learning Blog — 51
• The Guardian — 49
• CNews — 32
• AI News — 29

A blend of international business press (Bloomberg, Guardian), specialized tech media (TechCrunch, The Next Web), and Russian-language platforms (Habr, 3DNews, CNews) is deliberate. We gather the agenda from two markets at once and translate it into six languages.

Monthly Dynamics

Feed volume by month: January — 604, February — 1317, March — 164, April — 1777, May — 2019 (peak), June — 439, July — 1414.

Frankly: the gaps in March and June are technical pauses in our pipeline, not a decline in global news. The index reflects what passed through our editorial feed, not the absolute volume of the AI agenda in the world. The robust signals here are not the monthly surge, but the shares: who's in the headlines, which topics, which sources.

Methodology

The index is built on Hamidun News's own database: 7734 approved materials published from January 12 to July 19, 2026. Company mentions are substring searches by name in Russian-language headlines; isolated false matches are possible but do not change the order of magnitude. Topics are based on our categorization at the processing stage. Sources are by the source field of the original news. Live aggregates are updated on the [portal's analytics page](https://news.hamidun.com/ru/research).

How often is the index published?

This is the first edition; it covers the entire first half-year since the portal's launch — from January 12 to July 19, 2026. Going forward, we plan to publish the index monthly: with fresh shares of vendors, topics, and sources, so you can track shifts in the agenda from month to month.

Can these figures be cited?

Yes. All figures are taken from the Hamidun News database, and the period and methodology are indicated above. When citing, please specify the source: Hamidun News, news.hamidun.com. If you need a slice by a specific topic, company, or period — write to us, we'll calculate and send it.

Why do agents, not models, set the agenda?

Because the market has moved from the question "whose model is smarter" to "what can it do on its own." 892 headlines about agents versus 768 about OpenAI over the half-year — that's the answer: value has shifted from the model itself to what it autonomously does for humans.