AWS re:Invent 2026 conference guide
This guide provides ongoing coverage of AWS re:Invent 2026, including major announcements, trends, and product releases. The conference, scheduled from November 30 to December 4 in Las Vegas, will feature over 2,200 sessions on topics like AI, cloud computing, and security, along with five keynotes, including one from AWS CEO Matt Garman. The guide also offers a recap of AWS updates since the previous year's event.
Windows Server 2028 preview focuses on resilience, AI, security
Microsoft has released an initial preview of Windows Server 2028, emphasizing enhanced resilience, AI integration, and security features. The update aims to enable enterprises to handle AI workloads on-premises, with features like extended Resilient File System (ReFS) for boot volumes and quick machine recovery to improve system uptime. Additionally, the new OS offers native accelerator support for AI inferencing, Intune integration for server management, and advanced security measures including confidential VMs and post-quantum cryptography.
What BMC’s new mainframe chief learned about AI from IBM - HashiCorp
Tarun Chopra, BMC's new senior vice president and general manager of Intelligent Z Optimization and Transformation, advocates bringing AI to mainframes rather than trying to make mainframes AI-native. Having spent 24 years at IBM, Chopra believes the most significant ROI from AI on IBM Z lies in capturing mainframe knowledge for younger generations, leveraging insights from integrating HashiCorp tools with the Z platform. BMC's strategy, under Chopra's guidance, includes "agentifying the mainframe" and has launched BMC AMI Agentics to use AI to assist IT operators with mainframe issues and optimization.
The evolution of object storage for AI
Hitachi Vantara has introduced significant enhancements to its VSP One Object storage platform, increasing performance and supporting open-format data queries, to meet the escalating demands of AI environments. This move reflects a broader industry trend where object storage must evolve beyond cold archives to provide higher transactional performance and improved data management capabilities for AI initiatives. Organizations are realizing that existing storage infrastructures are insufficient for AI, necessitating upgrades to maximize ROI on AI investments.
AI agents execute transactions — but who authorizes them?
As AI agents begin to execute financial transactions, CIOs face critical challenges in establishing authorization and accountability. Companies like Visa, Bill, and Billtrust are piloting AI-executed payments, prompting the need for clear controls that define what agents can do, when human approval is required, and who is responsible for errors. This involves distinguishing between an agent's access and its authority, implementing authorization gateways, and negotiating clear responsibilities with AI vendors.
The rise of the AI decision model
Decision models are emerging as a crucial component in enterprise AI, designed for high-speed, low-cost judgments by selecting from predefined options with assigned probabilities. Unlike large language models, they excel at narrow, structured questions, making them ideal for tasks like ticket routing or invoice matching. This technology addresses limitations of traditional ML classifiers and LLMs for specific business processes, offering benefits in cost, speed, and ease of use, though challenges in accuracy and integration still require careful consideration.
The $0 cybersecurity initiative: 7 ways to improve security this week
This article outlines a seven-day, zero-cost cybersecurity initiative designed to help organizations improve security without increasing their budget. The plan focuses on leveraging existing capabilities through better deployment, configuration, and testing, addressing execution gaps rather than just capability gaps. It provides a daily checklist of actions such as enforcing MFA, removing excessive access, reducing attack surfaces, reviewing exceptions, verifying backups, documenting recovery paths, and establishing evidence of control effectiveness.
Microsoft launches new Postgres database, adds to Fabric capabilities
Microsoft has launched Azure HorizonDB, a new fully managed PostgreSQL database service, and Fabric HorizonDB, an optimized version for its Fabric platform. These new offerings are designed for the AI era, enabling AI agents to use live transactions and historical records in real-time. While competitors offer similar PostgreSQL capabilities, Microsoft's strength lies in its tight integration with its broader data and AI ecosystem, including OneLake, Power BI, Fabric Apps, and Copilot.
AI vendor exit strategies: Why you need a plan before you exit
Enterprises need robust exit strategies for AI vendors due to high dependency costs and technical challenges associated with switching providers. Unlike traditional SaaS lock-in, AI migrations involve complex considerations like prompt libraries, model behavior, and employee training. A proactive exit plan, including assessing costs, identifying dependencies, evaluating alternatives, and testing migration, helps minimize disruptions from price changes, model deprecation, regulatory shifts, or service issues.
Java4Shell: Critical Java runtime bug discovered ‘in one hour’ with AI
A critical memory-corruption vulnerability named Java4Shell (CVE-2026-114938) was discovered in OpenJDK's Base64 decoder by a researcher using AI in just one hour, costing less than $200. This flaw, scoring 9.2 on CVSS 4.0, allows attackers to leak sensitive heap data or execute arbitrary code in enterprise Java applications like Atlassian Jira and Jenkins. Organizations are urged to apply coordinated patches released by Oracle and the OpenJDK Vulnerability Group and implement temporary controls if immediate patching is not feasible.
Few payers are ready for looming electronic prior authorization deadline
A new survey by the Workgroup for Electronic Data Interchange (WEDI) reveals that only 1 in 10 payers are fully prepared to implement the electronic prior authorization requirements mandated by a 2024 regulation, despite a January 1 deadline. Payers are facing significant challenges with system connectivity, workflow adaptation, and higher-than-anticipated costs for compliance with the API requirements. While most payers are at least halfway there, providers are also largely unprepared, highlighting the complexity and cost of the new interoperability rules.
Starburst intros Context Studio to boost trust in AI responses
Starburst Data has introduced Context Studio, a new capability designed to improve the accuracy and trustworthiness of AI agent responses by integrating context from disparate systems. Unlike competitors that centralize data, Context Studio uses a federated architecture to collect and apply business logic and definitions without requiring data movement. This approach aims to address the common problem of AI initiatives failing due to agents lacking relevant and trusted contextual information.
Hybrid cloud helps energy company deploy AI amid SAP S/4HANA migration
Uniper, a German energy provider, is leveraging a hybrid cloud strategy to deploy AI capabilities, including a custom SAP Joule agent for procurement, while in the midst of migrating its legacy on-premises SAP ERP systems to SAP S/4HANA Cloud Private edition. This approach allows the company to benefit from AI-driven digital transformation ahead of its full S/4HANA implementation, integrating with Microsoft Azure and Microsoft 365. The company faces challenges in user adoption, public cloud limitations for specialized software, and navigating AI governance, but aims to overcome these by promoting a shift from fear to embracing AI's potential.
Microsoft unveils AI models to compete on cost and performance
Microsoft has launched new AI models, including Decision 1.1 and an expanded MAI family (MAI-Code-2, MAI-Image-2.7, MAI-Cyber-2-Flash), designed to compete on cost and performance against rivals like OpenAI and Anthropic. These models target enterprise challenges such as coding, image generation, and cybersecurity, offering significantly lower pricing compared to competitors. The company is also emphasizing its model router in Foundry to help enterprises automatically select the best model for specific use cases.
Zendesk expands its outcomes-based pricing and product strategy
Zendesk is expanding its outcomes-based pricing and product strategy, introducing new platform capabilities and flexible pricing models. The company is testing plans that move away from per-seat licenses to an "outcome budget" for resolved customer issues, coupled with new AI features like the Open Service Directory and Zendesk Bridge. This shift aims to differentiate between a simple "resolution" and a positive customer outcome, ensuring billing aligns with actual customer satisfaction.
ThoughtSpot debuts Customer Hosted to address data sovereignty
ThoughtSpot has launched Customer Hosted, a new deployment option for its data and AI platform, allowing enterprises to deploy on-premises and in private clouds (AWS and Google Cloud). This addresses critical data sovereignty, compliance, governance, and security concerns, especially amplified by the increasing use of AI agents. The offering brings ThoughtSpot to parity with competitors like Cloudera, Tableau, and Qlik, who already provide similar self-managed deployment options.
AI labs and ransomware: Who’s responsible for catching misuse?
A recently exposed ransomware operation utilizing commercial AI models has sparked debate over the responsibility of AI labs in detecting and flagging misuse of their technology. While AI is lowering the barrier to entry for ransomware attacks, making them more scalable and effective, the question remains whether the developers of these AI models should be held accountable for criminal exploitation or if the onus is on third-party security researchers to uncover such activities. AI labs are implementing self-monitoring and guardrails, but the legal framework around this responsibility is still developing.
Digital twins: The key to safer infrastructure modifications
Digital twins offer a critical solution for safer infrastructure modifications by simulating changes before deployment, allowing teams to assess potential impacts on interconnected systems. This approach helps identify and prevent disruptions that traditional methods like IaC or static topology maps might miss. While digital twins enhance validation and reduce risks, their effectiveness relies on accurate, up-to-date models and should be integrated with other robust testing and monitoring procedures.
What buying party balloons taught me about buying agentic AI
The author draws a parallel between purchasing party balloons and procuring agentic AI services, highlighting three pricing models: flat-rate, consumption-based, and outcome-based. He explains how different AI vendors, like SaaS companies and business process outsourcers, adapt these models and emphasizes the importance for enterprise buyers to understand their leverage points in contract negotiations, especially concerning the definition of "outcome" and metering. The article advises focusing negotiation efforts on these crucial details to ensure value and predictable costs.
New Qlik Answers capabilities improve AI-fueled insight generation
Qlik has introduced custom skills and advanced analytics capabilities for its Qlik Answers platform to enhance AI-fueled insight generation. These new features aim to provide the AI assistant with greater context by allowing enterprises to define business rules with their own terminology and enabling connections to broader datasets for predictive analysis. While these capabilities help Qlik remain competitive, analysts note that the company's multi-platform data management and governance are its key differentiators.
Tech leaders call for sharper focus on AI’s business value in 2027
Technology leaders are emphasizing a shift in AI focus for 2027, moving beyond mere technological capability to concentrate on tangible business value. Priorities include redesigning workflows, ensuring measurable outcomes, and building flexible architectures that can adapt to changing AI models. Experts also highlight the importance of assessing the cost of validating AI output and empowering managers on the ground to drive AI innovation.
Nvidia creates committee to oversee national security, chip exports
Nvidia has established a board-level committee to oversee national security policies, export-control compliance, and global trade risks, giving directors direct oversight amid tightening U.S. restrictions on advanced AI computing access for China. This move follows allegations of advanced Nvidia chips being diverted to China and increased scrutiny from Washington regarding the company's export practices. The committee aims to strengthen governance and engagement with government stakeholders, though analysts question its ultimate authority to compel business practice changes.
AWS databases chief on how data infrastructure powers AI agents
The AWS databases chief discusses how data infrastructure is crucial for powering AI agents, emphasizing the company's strategy of using specialized non-relational databases rather than a single general-purpose system. The conversation highlights the importance of data governance, accessibility, context, and proximity for AI agents to act on trustworthy information. AWS is enhancing its existing databases with AI-specific capabilities like vector indexing while maintaining consistent governance across its offerings.
Google finds AI agents autonomously turning patches into exploits
Google Threat Intelligence Group (GTIG) reports that threat actors are now using large language models (LLMs) and multi-agent frameworks to autonomously discover and exploit vulnerabilities, eliminating the need for human intervention. This development marks a "critical inflection point" in adversarial AI, significantly reducing the window between patch disclosure and weaponized exploitation. Security experts warn that traditional patching cycles are becoming obsolete, urging enterprises to adopt AI-enabled defense mechanisms to counter machine-speed attacks.
Silicon Valley startup showcases humanoid robot with human neurons
OpenMind, a Silicon Valley startup, has unveiled Iris, a humanoid robot that incorporates lab-grown human neurons to enhance its decision-making capabilities. These neurons, derived from stem cells, work alongside conventional computing to help the robot interpret its environment, understand social cues, and even store basic memories and influence its "mood." While still in early research stages, this development demonstrates the potential for neural cultures to integrate into robotic control loops, offering new avenues for learning, adaptation, and more energy-efficient, context-sensitive decision-making in robots.
CMS gives states another 6 months to implement Medicaid work requirements
The CMS has granted states a six-month delay, until June 30, to implement Medicaid work requirements, allowing more time to prepare systems for eligibility checks. States can apply for this "hold harmless" option by December 11, extending the original January 1 deadline. This decision comes after states struggled with IT system readiness and pushed back against the requirements, with some industry groups praising the delay to prevent beneficiaries from losing coverage due to administrative issues.
2027 AI outlook: Cheaper tokens, steeper AI bills
In 2027, the AI market will see decreasing costs for individual AI tokens due to a price war among major model makers and increased adoption of open-weight models. However, overall enterprise spending on AI platforms and models is predicted to increase significantly, rising to $99 billion, as companies accelerate AI usage and prioritize the quality of task execution over token cost. The market will also focus on better agent management, multi-model orchestration, and the expansion of robotics and physical AI, particularly mobile robots and cobots, while humanoids remain more aspirational than practical.
Another new CEO starts at New Relic, looks to beef up agentic features
Nandini Ramani, formerly of AWS, has taken over as CEO of New Relic, aiming to lead the company into an "agentic era" focusing on agent-driven observability features. She faces the challenge of differentiating New Relic in a highly competitive market, emphasizing a lean innovation agenda and an extensible headless architecture for agents. The company also needs to address customer concerns around pricing and vendor lock-in, while supporting open standards like OpenTelemetry.
AI agents are deleting databases. CISOs need a recovery plan now
Amid a rise in AI-driven failures, including agents deleting production environments and data, CISOs must prioritize resilience plans over mere guardrails. Forrester advises integrating AI risks into existing enterprise risk frameworks and developing comprehensive recovery capabilities like workarounds, rollback mechanisms, and kill switches. The focus is on preparing for inevitable AI failures by strengthening fundamental operational resilience.
What CIOs should learn from the Microsoft outage
The recent widespread Microsoft 365 outage exposed significant hidden cloud dependencies, prompting a visceral reaction from clients who want to remove Microsoft from their ecosystems. CIOs need to move beyond panic and focus on understanding their true vendor risk by mapping direct and indirect cloud dependencies. This requires building a comprehensive resilience plan and changing operating practices, including procurement, architecture reviews, communications, identity, and backup strategies, to ensure business continuity during future outages.
How does data traceability shape architecture choices?
Enterprises building modern data architectures face increasing complexity, especially with the rise of AI, which makes tracking data issues challenging. This article explores how data traceability can inform architectural decisions, emphasizing the need to reconstruct the entire data lifecycle, understand data provenance, and establish robust access controls. It also highlights the importance of evaluating vendor tools for their ability to provide end-to-end lineage and impact analysis, acknowledging that a combination of tools will likely be necessary to achieve comprehensive traceability.
Observability is becoming its own data swamp
Many enterprise observability initiatives, despite good intentions and significant data collection, are failing to provide real insight when systems break, becoming data swamps. The author argues that collecting massive amounts of telemetry without a clear strategy for action leads to overwhelming noise, alert fatigue, and increased costs, rather than faster problem resolution. Effective observability requires focusing on critical business outcomes, intentionally designing what data to collect, and ensuring signals lead to actionable insights.
When employees leave, their AI access may not
A survey by Sagiss found that many former employees still have access to company AI tools and data after leaving, posing significant security risks. The issue stems from a lack of coordination between HR, managers, and IT during offboarding, as well as the proliferation of personal and shadow AI tools. CIOs need to inventory AI usage, ensure comprehensive account termination, and verify that company data stored in AI tools is no longer accessible to former staff.
The rise of GreenOps: Using AI to power more sustainable IT
As IT operations expand, the environmental impact of data centers becomes a critical concern. GreenOps leverages AI to optimize IT for sustainability, reducing resource consumption, waste, and costs. This article explores how AI transforms traditional GreenOps into GreenAIOps, detailing key use cases, implementation challenges, and the business case for a more sustainable IT future.
5 data management and analytics trends to watch for in 2027
Data management and analytics in 2027 will be shaped by the increasing demands for real-time, context-rich data, proactive governance, and effective cost control, especially with the rise of agentic AI. Key trends include the further decentralization of data, the growing importance of ontology layers for providing context to AI agents, a shift towards proactive ("shift-left") data governance, and intensified focus on managing data costs as AI deployments scale. These trends collectively emphasize a move towards more efficient, secure, and intelligent data architectures.
10 reasons enterprise AI success hinges on platform teams in 2027
As enterprises increasingly adopt AI, platform teams will become crucial in managing distributed AI infrastructure, inference, risks, and costs by 2027. This shift involves consolidating heterogeneous agents, optimizing workload routing across diverse environments, and addressing challenges like hardware shortages, skills gaps, and escalating token costs. Centralized platform engineering will be essential for ensuring AI governance, security, and financial viability, particularly as AI inference workloads become more distributed and complex.
Cloudera data management comes through for comms tech provider
Amdocs, a communications technology provider, adopted Cloudera's data management platform in 2014 to handle its rapidly increasing data volume, moving away from expensive and less scalable Oracle databases. Despite a challenging two-year transition, the partnership enabled Amdocs to effectively manage complex data, reduce total cost of ownership, and develop data products for its customers. Amdocs is now exploring using Cloudera's hybrid platform for agentic AI initiatives, recognizing its value for on-premises solutions amidst rising cloud costs.
5 AI trends CIOs should watch in 2027
In 2027, CIOs are poised to intensify their focus on AI investments, shifting from numerous pilot projects to a select few high-impact initiatives. Key trends include redesigning business processes to enhance customer experience, carefully selecting AI models based on task requirements, involving finance and business leaders in spending decisions, and formalizing "AI champion" networks to boost employee adoption. These strategies aim to move AI from experimental stages to widespread, consistent implementation across enterprises.
Shadow AI: Defending against unsanctioned browser-based agents
Shadow AI, particularly through browser-based agents, poses significant security risks by circumventing IT controls, exposing sensitive data, and automating risky tasks without oversight. A 2026 Microsoft Cyber Pulse report indicates that at least one-third of employees use unauthorized AI tools, highlighting the widespread nature of this threat. To counter shadow AI, organizations need a multi-layered security strategy that includes real-time visibility into IT assets, robust identity and access management for AI agents, contextual data loss prevention, and continuous employee education on potential dangers.
The Prompt: Nvidia’s rivals have an opening, but it’s unclear if they can deliver
Nvidia's competitors, AMD and Google, have introduced new AI hardware, creating an opportunity to challenge Nvidia's long-standing dominance, especially after a leadership shakeup at Nvidia. While these new products offer alternatives, convincing enterprise customers to switch will be difficult due to existing investments in Nvidia's software ecosystem and infrastructure. Although Nvidia's internal issues might prompt some reconsideration, a significant shift in market share for rivals is not yet evident, as switching involves more than just chip replacement.
Once more, HubSpot swaps CEOs
HubSpot has once again changed CEOs, with co-founder Brian Halligan returning to replace Yamini Rangan, who stepped down as the company's stock has lost significant value. This leadership shuffle, dubbed a "refounding" by Halligan, comes amidst several controversies including recent layoffs, proposed changes to terms of service, and a whistleblower lawsuit. Despite these issues, the company has seen growth in its customer base and revenue under Rangan's tenure, but faces challenges in demonstrating ROI for its new AI agents and maintaining customer experience during the transition.
Hyland opens enterprise content to AI agents with OpenArch
Hyland has launched OpenArch, an open-source content index designed to connect enterprise content from platforms like Alfresco and Nuxeo to AI services and agents. The project, led by open-source veteran John Newton, aims to preserve metadata, permissions, and business rules when content is accessed by AI, ensuring governance and compliance. OpenArch allows enterprises to provide AI agents with a unified view of scattered content while maintaining control over data security and permissions.
Anthropic IPO tests Wall Street’s appetite for standalone frontier AI
Anthropic's upcoming IPO will assess Wall Street's confidence in standalone AI companies, contrasting their business models with diversified tech giants. The offering will also shed light on the financial risks, significant infrastructure commitments, and potential implications for enterprise customers. Investors will closely examine Anthropic's ability to convert rapid revenue growth into a sustainable, profitable business despite substantial operating costs and long-term infrastructure liabilities.
Jensen Huang’s scandal-fueled ouster exposes enterprise doubts about AI
Nvidia's founder and CEO, Jensen Huang, has resigned due to alleged extramarital affairs, including one with a suspected Chinese spy. While the departure might not immediately harm Nvidia's dominant position in AI chip production, it highlights growing skepticism among enterprises regarding the unclear return on investment from AI technology. The incident could further fuel doubts and potentially lead enterprises to seek alternative solutions beyond Nvidia.
AI cost management isn’t rocket science
Forrester principal analyst Kevin Ogunsua states that managing AI spending is achievable through clear budgets, vendor visibility, and financial discipline. He emphasizes the unpredictable nature of AI agent costs due to variable model calls and agent interactions, and outlines a framework focusing on bounding requests, reducing calls, and setting limits. Ogunsua advises CIOs to demand observability from vendors, establish handlers for budget overruns, and engage CFOs and procurement to adjust contracts for outcome-based AI investments.
Agility Robotics first U.S. specialist humanoid maker to go public
Agility Robotics has become the first publicly listed U.S. humanoid robotics company, trading on Nasdaq under the ticker AGLT after a business combination. The company raised approximately $465 million, which it plans to use to accelerate the commercialization of its Digit humanoid robot, particularly for manufacturing and logistics applications. Analysts view this IPO as significant for the U.S. robotics market and a reflection of growing confidence in the commercial potential of robotics technology, despite challenges in scaling deployments in heavily automated environments.
Deep dive: The state of AI adoption in 2026
This deep dive analyzes the state of AI adoption in 2026, using governmental and private data to illustrate current trends and predict future developments for 2027. Key findings highlight a scaling bottleneck for AI technologies, particularly agentic AI, and a stagnation in profitability gains across sectors, with only about a third of organizations reporting positive financial returns attributed to AI. The article concludes that while AI has found a foothold, its progress may plateau without more specialized tooling, suggesting most organizations are not significantly lagging behind.
The future of data and analytics lies in decision intelligence
Enterprises have invested heavily in data platforms and dashboards, but turning that data into fast, trusted decisions remains challenging. Shayde Christian, chief data and analytics officer at Cloudera, argues that AI-driven decision intelligence, built on strong governance and security, can bridge this gap. This approach uses a "single pane of glass" to provide contextualized insights, significantly reducing the time to insight and enabling proactive, automated business actions.
With IPO looming, Anthropic launches new AI models, workplace tools
Anthropic is preparing for its IPO with a projected valuation of $1.33 trillion, well below earlier expectations. The company has released updated AI models, Claude Fable 5.2 and Claude Mythos 5.2, focusing on improvements in coding, reasoning, and vision, along with new Claude Enterprise capabilities for various workplace tasks. Despite market upheaval, the public offering aims to provide greater financial transparency and access to capital for the generative AI vendor.
Hospitals write off $132B as AI fails to move the needle with finances
Hospitals experienced an estimated $132 billion in write-offs last year due to bad debt and final claim denials, an 18% increase from 2024, despite growing investments in revenue cycle technology, including AI tools. The analysis suggests that while AI streamlines processes, it may not be providing the in-depth data analysis needed to optimize financial performance or address structural issues like prior authorizations. This puts hospitals in an "AI arms race" with payers, requiring strategic technology investments to remain competitive and improve key financial indicators.