It doesn’t look like it – This week has been a bit week for AI (as usual). The general trend being AI companies telling us that “AI is no longer a pilot or experiment” and is starting to add value to the “bottom line”. For businesses it is clear that vendor choice(s) made now will shape procurement, skilling, architecture, and governance for years to come.
Here’s my take on the top AI stories that I’ve read this week (thanks to Copilot for condensing my rambles)!
Big AI Stories this week – 01/05/26
Microsoft AI News
This week was all about Microsoft Copilot Adoption at scale and a new commercial agreement with OpenAI
In this week’s earning report, Microsoft said that paid Microsoft 365 Copilot seats surpassed 20 million which is up from 15 million in just 3 months. They also reported an AI annual run‑rate of $37B which is up ~123% YoY.
“Copilot queries per user were up nearly 20% quarter‑over‑quarter…weekly engagement is now at the same level as Outlook“.
Satya nadella | CEO | microsoft
Also, this week, Microsoft and OpenAI “amended” their partnership, so Microsoft keeps long‑term IP access (until 2032) while moving away from the old revenue‑share exclusivity. This gives Microsoft more direct monetisation flexibility.
Microsoft also went to GA with Microsoft 365 E7 (their “Frontier suite”), Agent 365 (the control plane for agents) and Agent Mode (“Edit with”) agentic update to Copilot across the Office Apps… Big big updates.
This is proof that enterprise AI can be a major P&L for AI companies. As adoption increases, I expect to see vendors favour “usage/consumption based” billing, tighter Office and App integration, and a renewed focus on IP and lock‑in.
The Official Microsoft Blog
Cisco AI News
Whilst not an AI Platform provider as such, Cisco continue to position themselves as the infrastructure what runs AI. They have unveiled a new push for AI‑native networking which includes the Silicon One G300 switch ASIC, new N9000/8000 systems, a Nexus One unified management plane, and a big update to their AI Defense which provides AI model‑level security. Cisco frame the network as part of compute to reduce GPU idle time and speed job completion and to ensure AI traffic can flow as it needs which is more east to west then north to south.
This is important for any organisation that runs or connect to large model clusters since, network design, cooling, and observability now materially affect cost and throughput. Cisco Newsroom Cisco Newsroom
Apple AI News
After a few Siri failures, Apple now has announced that Tim Cook will become Executive Chairman and John Ternus will be CEO from September 1 2026
Separately, Apple is said to be preparing a major Siri overhaul (iOS 27 / WWDC) with a chatbot‑style interface and deeper on‑device AI with AI everywhere.
This will likely mean (according to Apple Gadgets 360) new device hardware, management, SDKs, and privacy/edge tradeoffs for enterprise apps.
Adobe AI News
Adobe have launched Firefly AI Assistant (into public beta). This is a conversational agent that orchestrates multi‑step workflows across Photoshop, Premiere, Express and more from one prompt – Think Copilot or Claude Cowork!
Adobe has positioned this as “agentic creativity” where the assistant executes while creators stay in control with multi-step context aware working.
OpenAI + Google AI News
This week has also seen OpenAI release GPT‑5.5 (already rolling out in M365 Copilot too) which has been “said” to be a big step toward agentic work and stronger at coding and multi‑step tasks – hot on the heals of Claude.
OpenAI framed it as “our smartest and most intuitive‑to‑use model yet.”.
Google is also moving really fast as they tend to do when they are in win and acquire share mode. This week they committed to up to $40B to Anthropic and have also signed agreements to let Gemini be used on classified Pentagon networks.
We may hear about the AI bubble burst, but it just seems to bet getting bigger. These are more huge investments.
It’s a Multi-Modal Future
As AI and AI-powered tools with agentic capabilities develop and improve on scale, it’s important to determine where the human vs machine (well agent) plays.
Many SaaS leads and Model Creators are working with other model providers meaning we are in a multi-modal world. These have different costs, operating regions, terms and conditions. As such businesses need to look at Auditing and control tools like Microsoft Agent 365, Cisco AI Defense etc and ensure there is solid logging, controls and cost monitoring. Human-in-the-Loop and clear guidance for people over agentic AI is also critical.
Microsoft have an interesting approach – expanding with Anthropic and Open AI while they build and expand their own first-party models
Leave a Reply