The latest AI news from OpenAI, Anthropic, AWS, Google, Microsoft, Cohere, DeepSeek, sovereign AI efforts, and an agentic AI customer reference story from Jernia.
AI Transformation Brief
VOL 3 MAY 2026
Featuring AI best practices and insights to enable our members to strategize, plan, develop, deploy, manage, and govern AI-based technologies and solutions.
In This Issue –› AI in the News | AI Research Highlights | Vendor Spotlight | Upcoming Events & Resources
AI IN THE NEWS
OpenAI exclusivity ends with Microsoft, OpenAI models arrive on Amazon Bedrock, and GPT-5.5-Cyber released
OpenAI exclusivity with Microsoft ends
OpenAI portfolio now available on Amazon
On May 7, 2026, GPT‑5.5‑Cyber was made available in limited preview to defenders responsible for securing critical infrastructure.
On April 28, 2026, OpenAI models arrive on Amazon Bedrock.
On April 27, 2026, Microsoft and OpenAI officially announced the end of their exclusivity agreement.
ANALYST ANALYSIS
The OpenAI’s exclusive relationship with Microsoft drove Azure's growth and gave Microsoft a huge head start in the AI industry. Ending this exclusivity allows OpenAI to pursue relationships with the other major cloud providers. A win for OpenAI.
Not surprisingly, a joint announcement with OpenAI and AWS was made the very next day about the availability of the OpenAI products on Amazon Bedrock.
GPT-5.5-Cyber is a specialized, "cyber-permissive" variant of GPT-5.5. It removes the standard defensive guardrails and enables researchers to reverse-engineer binaries, study live malware, and perform controlled penetration tests.
Anthropic revenue passes OpenAI for the first time, joint commitment with AWS to build a 5GW data center, and Claude Security beta released
Anthropic surpasses OpenAI in corporate spend
Anthropic commits a 5GW data center with AWS
On May 13, 2026, according to data from the Ramp AI Index, during the month of April, Anthropic surpassed OpenAI in business spend for the first time.
On April 30, 2026, Claude Security was made available to all Enterprise customers. This represents a trend of the AI model vendors focusing on improving cybersecurity operations.
On April 20, 2026, Anthropic announced a 5GW data center commitment with AWS.
ANALYST ANALYSIS
OpenAI has dominated enterprise AI since late 2022. Yet Anthropic now leads in corporate spending share (34.4% to 32.3%), proving that loyalty to any AI vendor is fleeting.
The ongoing announcements of multi-GW data centers demonstrates a concern that the ultimate bottlenecks will be energy and silicon. This latest nuclear-scale commitment also represents a departure from dependence on NVIDIA, by committing to AWS Trainium chips to address compute demands.
What makes Claude Security different is that it acts like a human expert, tracing data across files, analyzing business logic, and uncovering hidden exploit chains using Claude Opus 4.7.
INFO-TECH AI INSIGHTS
AI Transformation Brief
#2
AI IN THE NEWS
OpenAI now available on Amazon Bedrock, updated Amazon Quick available as a desktop application and Bedrock AgentCore Payments released
OpenAI models, Codex, and Managed Agents now available
Amazon Bedrock AgentCore Payments
On May 7, 2026, Amazon Bedrock AgentCore payments was released, positioned as the first managed end-to-end payment capabilities for agents.
On April 28, 2026, Amazon announced the availability of OpenAI models, Codex and Managed Agents on Amazon Bedrock.
Also on April 28, 2026, Amazon Quick is now available Amazon Quick now available as a desktop application for macOS and Windows.
ANALYST ANALYSIS
The availability and integration of OpenAI’s products on Amazon Bedrock is a massive shift in the cloud landscape, breaking the Microsoft exclusivity with OpenAI, representing a huge win for Amazon.
The availability of Amazon Quick on the desktop can be viewed as Amazon’s version of a safe and enterprise-grade version of the capabilities that popularized OpenClaw, a desktop agent that became wildly popular after its introduction.
Bedrock AgentCore Payments provides a secure architecture that enables AI agents to perform financial transactions, manage budgets, and handle procurement autonomously.
Google announces Gemini Intelligence, Googlebook, and Gemini Enterprise Agent Platform
Gemini Intelligence and Googlebook
Gemini Enterprise Agent Platform
On May 19, 2026, Google launched Gemini Omni at I/O 2026 – a "world model" family that understands physics and motion for AI video generation and prompt-based video editing.
On May 12, 2026, Google introduced Gemini Intelligence for Android alongside its new "Googlebook" laptop category.
On April 22, 2026, Google announced Gemini Enterprise Agent Platform, a massive evolution of Vertex AI engineered to build, scale, and govern fleets of fully autonomous AI agents.
ANALYST ANALYSIS
The integration of Gemini into the operating system, the unveiling of Gemini Intelligence, and the new Googlebook laptop line, represent a direct competitive threat against Apple’s and Microsoft’s desktop ecosystems.
Google’s Agent Platform introduces much-needed governance architectures like Agent Identity (assigns every AI agent a unique cryptographic ID) and Agent Gateway ("air traffic control" for multi-agent workflows) to provide improved auditable and secure resources.
Gemini Omni introduces the reasoning capabilities of Google DeepMind's previous separate models (like Veo and Genie), the new system delivers a mathematically accurate grasp of real-world physics, kinetic energy, and fluid dynamics. This will transform video generation from simple prompts to highly structured, conversational editing dialogues.
Microsoft/OpenAI exclusivity ends, announces MDASH and AI pricing model changes
Microsoft Multi-Model Agentic Security System
Microsoft AI pricing model change to consumption-based
Microsoft OpenAI Exclusivity Ends
On May 12, Microsoft introduced its new multi-model agentic security system (MDASH).
On April 27, 2026, Microsoft announced it was introducing a new pricing model for all its AI-enhanced products. Starting with GitHub Copilot, transitioning it from seat-based pricing to a consumption-based model.
On April 27, 2026, OpenAI and Microsoft ended their exclusive relationship.
ANALYST ANALYSIS
Microsoft’s MDASH represents a shift from simple, single-prompt coding assistants to complex, autonomous cybersecurity ecosystems. Instead of depending on a single large language model, MDASH orchestrates over 100 specialized AI agents that span multiple frontier and distilled models.
The change in pricing of any AI-augmented offering to a token-based or a consumption-based model addresses the increasing infrastructure costs while only charging a flat-rate for compute services.
With the end of the exclusivity of OpenAI products available on Azure only, the relationship with OpenAI enters a new phase:
- New OpenAI products will be made available first on Azure unless Microsoft cannot or chooses not to support them.
- Microsoft will continue to have a license to OpenAI IP for models and products through 2032.
- Microsoft will no longer pay a revenue share to OpenAI.
INFO-TECH AI INSIGHTS
AI Transformation Brief
#3
AI IN THE NEWS
Cohere announces Command A+ their premier open source and open weights model and embraces sovereign AI initiatives
On May 20, 2026, Cohere announced the availability of its most capable AI model, Command A+. The co-founders of Cohere have explicitly framed this release as their first fully open-source flagship model, explicitly leaning into the Apache 2.0 framework to provide a predictable, compliant engine for sovereign enterprise infrastructure.
ANALYST ANALYSIS
The release of Command A+ is a strategic milestone for Cohere, it represents their latest and most capable AI model, leveraging a mixture of experts (MoE) architecture and optimized for complex, multimodal enterprise workloads. It is also embracing the global “Sovereign AI” movement by releasing the weights under an open Apache 2.0 license. This release targets regulated industries, public sector agencies, and allied nations that need complete data autonomy. The model can run entirely on-premises, in private clouds, or in air-gapped sandboxes – offering critical infrastructure defenders an alternative to closed hyperscalers, OpenAI, and Anthropic. Finally, the open weights model reduces vendor lock-in while ensuring compliance with regulatory frameworks like the EU AI Act.
DeepSeek-V4 released and available on NVIDIA and Huawei
On April 24, DeepSeek V4 was officially released in public preview.
DeepSeek-V4-Flash: A smaller, rapid-inference model featuring 284 billion total parameters (13 billion activated per token). DeepSeek-V4-Pro: The flagship heavy-reasoning model featuring 1.6 trillion total parameters (49 billion activated per token).
ANALYST ANALYSIS
DeepSeek-V4-Flash is built as an ultra-economical, high-throughput model for standard completions and chat, while the flagship V4-Pro – a 1.6T MoE model with 49 billion active parameters – is optimized for deeper reasoning and complex software engineering tasks; both models support the same configurable triple-mode reasoning architecture ("Non-Think," "Think High," and "Think Max"), using specialized chain-of-thought processing for complex problem-solving and the flagship V4-Pro showcases massive gains in autonomous capabilities.
The most surprising capability is that DeepSeek-V4 is available on both NVIDIA and Huawei platforms, offering organizations a choice and breaks its dependency on NVIDIA chip technology and the CUDA framework. Organizations can select hardware that is best for their geography and enables the AI models to be independent of the underlying chip technology.
Jernia leverages agentic AI to improve the customer experience
On January 15, 2026, Jernia, Norway's leading hardware retailer, had finished deploying an agentic AI customer service solution using natural language and direct tool calling – resulting in faster responses, lower ticket volumes, and improved dialogue quality.
ANALYST ANALYSIS
NeuroSYS, Digital Transformation/AI services firm built a customized AI chat solution as a managed service. The application leverages Jernia’s own knowledge base to align the responses to ensure accuracy and brand control.
The system improves the customer experience with these capabilities:
- Answers questions about opening hours, store locations, terms and returns
- Retrieves live order status via nShift (integration software)
- Provides product tips and troubleshooting advice
- Runs on a dedicated instance with Jernia's own documentation
During the PoC, the agent resolved 46% of meaningful conversations without human intervention. Customers received faster responses, the support team handled fewer routine tickets, and the overall experience improved – all without increasing headcount.
INFO-TECH AI INSIGHTS
AI Transformation Brief
#4
AI RESEARCH HIGHLIGHTS
RESEARCH SPOTLIGHT:
Activate Data Governance With a Decision-Focused Framework New!
Source: Activate Data Governance With a Decision-Focused Framework
Info-Tech Research Group
VENDOR SPOTLIGHT:
- Acquia Source: The Command Center Vision Anchored in Governance
- Most Enterprises Believe in Open Source. Almost None Can Operate It.
- Your Security Tools Were Built for People. Agents Are Not People
- Chata.ai: Deterministic AI That Does Not Lie
- Google Cloud Next 2026: It's All About the Agents
UPCOMING AND RECENT EVENTS
- LIVE Las Vegas – Las Vegas, Nevada,, June 9-11, 2026
- LIVE Barcelona – Barcelona, Spain,, September 22-23, 2026
- LIVE Toronto – Las Vegas, Nevada,, November 10-12, 2026
- LIVE Gold Coast – Las Vegas, Nevada,, March 16-17, 2027
AI AND DATA ANALYTICS SOLUTIONS – RESOURCES
- AI Marketplace
- Artificial Intelligence Research Center
- AI Workforce Development Program
- AI Workshops
- AI Strategy
- AI Governance, Liability & Risk
- Accelerate Adoption of Gen AI
- Drive Business Value With M365 Copilot
- Design Your Agentic AI Prototype
- Develop Your Agentic AI Prototype
- Scale and Optimize Your Agent Implementation
AI EDITOR-IN-CHIEF
Bill Wong – Info-Tech AI Research Fellow
