Bill Gates Urges Slower AI Development, Warns of Risks

ThinkingAI, which builds autonomous growth software for consumer and gaming companies, has opened its analytics, engagement and agent stack to game studios that buy software with a card instead of a procurement cycle, offering four self-serve plans priced from $0 to $999 a month. The Free tier costs $0 and includes 2 million events a month, 1 project, 1 seat, 12 months of data retention and $10 in Agent Credits for the first month, with no credit card required. Starter costs $199 a month for 5 million events, 2 projects, 5 seats, 24 months of retention and $25 a month in Agent Credits, while Pro costs $399 a month for 10 million events, 2 projects, 10 seats, 48 months of retention and $50 a month in Agent Credits. Team costs $999 a month for 30 million events, 3 projects, 15 seats, 60 months of retention and $100 a month in Agent Credits, and an Enterprise tier is sold by agreement with sales for larger studios with compliance or data residency requirements. Annual billing saves 20%, paid plans are billed through Stripe and can be cancelled at any time, and a studio that outgrows its event allowance moves up a plan inside the product with no re-instrumentation and no migration. The plans follow ThinkingAI's Sept. 16 launch of the Agentic Engine and draw on 10 years of behavioral data across more than 1,500 enterprises and 8,000 applications, including Sega, Krafton, Habby, Century Games, Xiaomi, Yostar, TCL and FunPlus, and are being announced at the ThinkingAI Agentic Game Growth Summit in Istanbul, co-hosted with Mobidictum.

Jim Cramer said on the September 30 episode of Mad Money that Microsoft's massive artificial intelligence investment is entering a new phase as the company begins turning its data-center buildout into higher revenue. Cramer noted Copilot now has 30 million users and growing, Azure is accelerating rather than contracting, and Microsoft is beginning to recoup its monster data center investment, which he compared to the payback period Nvidia's Jensen Huang once described for his semiconductors. In its fiscal fourth quarter, Microsoft Cloud revenue increased 27% to $59.3 billion, while Azure and other cloud services revenue grew 43%, and commercial remaining performance obligations rose 84% to $678 billion, including a 25% increase excluding OpenAI. CFO Amy Hood said on September 9 that Microsoft had reduced dock-to-live times for infrastructure by more than 50% over the year, even as capital expenditures reached $41 billion in the June quarter and first-quarter fiscal 2027 spending is expected to exceed $50 billion. Microsoft Cloud's gross margin fell to 65% in the June quarter, and the company generated $19.6 billion of free cash flow while paying $35.8 billion for property and equipment, leaving margins and cash flow worth watching as spending continues to rise.

Atlassian Corporation announced the launch of its Forward Deployed Engineering program, which embeds senior applied AI engineers directly into customer environments to deploy production-ready AI solutions. The program targets a gap between AI adoption and business impact: while 85% of knowledge workers use AI, only 6% of executives can point to clear, organization-wide ROI, according to Atlassian's report Leading with Context: Lessons from Atlassian's AI Journey. Avani Prabhakar, Chief People and AI Enablement Officer at Atlassian, said the company is seeing 50% higher AI adoption in engaged accounts as a result of embedding engineers with customers. In Atlassian testing, grounding AI agents in the Teamwork Graph produced answers that were 44% higher quality while using 48% fewer tokens. Expedia used the program to archive 600,000 pages across Confluence and save approximately 3,200 hours annually, with engineering teams now running 10,000+ agent invocations each month across 50 live projects, while Reddit is partnering with the engineers to accelerate its enterprise AI transformation. Engagements typically run 12 weeks, and Atlassian is also releasing The Atlassian Forward Deployed Engineering Playbook.

eDreams ODIGEO announced a strategic collaboration with Alexa+, Amazon's next-generation AI assistant, to power travel discovery and booking. The integration uses eDreams ODIGEO's proprietary Model Context Protocol technology to connect its travel platform and operational voice to Alexa+, letting customers find and book flights, hotels, and car rentals through natural-language conversation. The company said the deal opens a new acquisition channel as its travel subscription model expands toward a goal of 13 million subscribers by March 2030. eDreams ODIGEO Chief Technology Officer Carsten Bernhard said conversational artificial intelligence is the latest major technological wave and a powerful accelerator for the business. The collaboration follows the Alexa Plus Partner summit held in Seattle last July, where eDreams ODIGEO was featured in the CTO Innovation Series hosted by Amazon Alexa.

Atlassian announced AMP: The Agentic Multiplayer Protocol at its Team '26 Europe conference, a new protocol that lets AI agents work alongside human teams in the flow of work. AMP gives agents context, identity, and governance, allowing them to participate through @mentions in Confluence, Jira comment threads, or Loom video briefs, while the Atlassian MCP Server pulls in context from third-party tools like Figma or IDEs. Every agent under AMP has a clear owner and distinct profile, appearing in real-time presence bars and cursors alongside human teammates, and agents are grounded in the Teamwork Graph, which now connects more than 250 billion objects and relationships. Atlassian also introduced a new Atlassian MCP Server, which now has nearly 2 million monthly active users and handles over 15 million daily tool calls, a 15x jump in six months, using up to 25% fewer tokens for the same Jira and Confluence work in internal benchmarking on Claude models. The company said it has shipped over 120 enterprise capabilities during the last year, with people and agents now working together more than 10 million times a month, and announced new capabilities including Agent Sessions, Non-Human Identities, Record for Agent, and Interactive PR Reviews.

ServiceNow has launched Flow by ServiceNow, a conversational AI-native service desk that can be deployed in a day with no infrastructure, while partners Recolabs and LogicMonitor expanded integrations to extend security and AI agent capabilities on the ServiceNow AI Platform. The launches and partnerships come alongside client outcomes including INRY's rapid EmployeeWorks rollout and the Bundesliga's consolidated fan-service hub, showing ServiceNow's AI-centric workflows being adopted across IT, employee experience, and CRM-style fan engagement. ServiceNow's narrative projects $25.1 billion revenue and $4.1 billion earnings by 2029, requiring 19.4% yearly revenue growth and roughly a $2.4 billion earnings increase from $1.7 billion today, with a $144.99 fair value implying 5% upside to its current price. The lowest analyst estimates are more cautious, with revenue reaching only about US$24.5 billion and earnings about US$2.4 billion by 2029, and skepticism that ServiceNow can sustain 20 percent to 30 percent AI price uplifts. The Flow launch and new AI partnerships reinforce the core AI workflow thesis but do not materially change near-term catalysts around AI monetisation or the key risks tied to security incidents and potential multiple compression.