18 November · AI between workplace tools and infrastructure

This morning’s briefing brings together three updates dated 17 November: new models for professional work, coding tools and the financial results of the infrastructure behind them. The sources are the original OpenAI, Microsoft and NVIDIA announcements. Stated availability, performance claims and financial results are different kinds of information; keeping them separate helps explain what is actually changing.

OpenAI · GPT‑6.2 Astra and Sol arrive

On 17 November, OpenAI announced GPT‑6.2 Astra and Sol for ChatGPT Work, Codex and the API. Astra is positioned for complex professional projects, and Sol for everyday work and agent workflows. The company reports improvements in document production, coding and computer use. These are manufacturer assessments that should be compared with testing in users’ own work settings. The API changelog confirms both models and states that tool calling requires the Responses API.

OpenAI · GPT‑6.2 Astra and Sol, announcement dated 17 November 2026 ↗

Microsoft · Coding brings costs into focus

In its 17 November Ignite keynote, Microsoft introduced Code 2 and Code 2 Flash and said they were generally available in GitHub Copilot and Foundry. The first targets tasks including changes across multiple files and debugging; Flash focuses on fast, routine edits. The company also announced MAI Licensed Weights, providing selected customers with access to its model weights. This is not an unrestricted release under an open license. Value should be assessed across the whole task, including checks and corrections.

Microsoft · Ignite keynote, 17 November 2026 ↗

NVIDIA · Quarterly revenue reaches $115.8 billion

In its 17 November release, NVIDIA reported revenue of $115.8 billion for the third quarter of fiscal 2027, which ended on 25 October 2026, up 103% from a year earlier. Data Center revenue was $107.6 billion, up 110%. The company also reported that shipments of Vera Rubin NVL72 had begun. Its forecast of approximately $130 billion in revenue for the next quarter remains a company projection, distinct from reported results. The figures document infrastructure demand, not the profitability of every AI application.

NVIDIA · Third-quarter fiscal 2027 results, 17 November 2026 ↗

The takeaway · From models to useful results

Competition is unfolding on two levels: making workplace tools more accessible and supporting the infrastructure needed to run them. For adopters, the useful comparison concerns time saved, output quality, controls and total cost. A new model or rising revenue does not replace that assessment.

This edition’s social kit

Italian copy for Instagram, LinkedIn and X, plus Italian and English covers and social previews, are available to the team. No content has been sent to social platforms.

Download copy and graphics for 18 November ↗

Sources and documents

Document date or reference period: Overnight between 17 and 18 November 2026, including announcements dated 17 November.

Selection and summary prepared with AI assistance. Original sources checked on the morning of 18 November; the announcements are dated 17 November. Not all sources provide a publication time, so events without a verified time are not described as occurring overnight. Manufacturers’ results and assessments are attributed to them.

Report an update or correction →