InfoQ's 2026 AI and ML trends report highlights a shift toward smaller, more efficient models. Small language models and reasoning models are replacing larger ones as the dominant approach. AI agents are also moving from personal tools to enterprise-wide deployment, with new protocols like MCP and A2A gaining traction. Inference engines such as vLLM, SGLang, and llm-d are now central to the stack, and inference is expected to consume about two-thirds of AI compute spending by year end.
Nvidia has begun shipping RTX Spark laptops through partners including ASUS, Dell, HP, Lenovo, Microsoft, and MSI. These machines pair Blackwell RTX graphics with Grace CPUs, offering up to 128GB of unified memory and 1 petaflop of AI compute. Starting at USD $2,399, they target users who need AI processing on device rather than relying on cloud services. Nvidia held launch events in six cities to demonstrate local AI application performance.
On the infrastructure side, semiconductor reliability is emerging as a critical challenge for large AI clusters. Building systems with 10,000 GPUs demands advances in PCBs, advanced packaging, thermal management, and power delivery. The focus is shifting from GPUs and high-bandwidth memory to the underlying engineering that keeps massive systems running without failure.
Enterprise adoption of AI is not without growing pains. A SecurityHQ and Datadog survey of 204 cybersecurity leaders found that while 76% have formal accountability rules, 45% have faced incidents where responsibility for AI agent actions was unclear. Only 49% could fully reconstruct what an AI agent did in the past 30 days, and just 17% could do so within minutes. Nearly half of leaders suspect unapproved AI agents are in use inside their organizations.
Human workers are bearing the brunt of the AI expansion in cybersecurity. Reports show AI-related cyberattacks have risen 300% in the past year, mostly launched by humans using AI tools. The average worker now spends over half their time on AI-related tasks, and many are experiencing burnout. The findings point to a need for better tools and investment to support the people on the front lines.
AI is also affecting how industries think about systems and talent. Semiconductor design is shifting from single chips to full packages, racks, and supply chains, with connectivity now a bigger limit than raw compute. In marketing, PepsiCo's Sam Wong said 85% of AI recommendations are accepted, but Jelmar's Volles stressed that human judgment remains essential for decisions that affect consumer trust. Together, these developments show AI is reshaping both technology and the people who work with it.
Key Takeaways
- InfoQ's 2026 report shows small language models and reasoning models replacing large models, with inference expected to account for about two-thirds of AI compute spending.
- Nvidia launched RTX Spark laptops with partners including Microsoft, Dell, and HP, offering up to 128GB unified memory and 1 petaflop of AI compute starting at USD $2,399.
- Semiconductor reliability is becoming a key focus for 10,000-GPU AI clusters, with PCBs, packaging, and thermal management now critical engineering challenges.
- A SecurityHQ and Datadog survey of 204 cybersecurity leaders found 45% faced unclear accountability for AI agent actions, and nearly half suspect unapproved AI agents are in use.
- AI-related cyberattacks have increased 300% in the past year, and the average worker now spends over 50% of their time on AI-related tasks, leading to burnout.
- AI semiconductor trends show connectivity is now a bigger performance limit than raw computing power, with development cycles shortened to six to nine months.
- At a Chicago marketing event, PepsiCo reported 85% of AI recommendations were accepted, while Jelmar emphasized the need for human judgment in consumer-facing decisions.
- Kolon Benit partnered with Dell Technologies Korea and Nutanix to train engineers on AI infrastructure including PowerStore Elite and PowerEdge 18G servers.
- Security experts urge vendors to treat AI-generated code in alarm monitoring software as unverified and require human confirmation before deployment.
- A survey of Gen Z Australians found 69% feel guilty when relaxing, with AI concerns rising to the fourth biggest worry for young adults.
InfoQ reveals key AI and ML trends for 2026
InfoQ published its 2026 trends report on AI, ML, and Data Engineering. Small language models and reasoning models are replacing large models. AI agents are moving from personal use to enterprise-wide deployment. The report also highlights AI evaluations, physical AI, and predictions for robotics and cheaper models in the coming year.
InfoQ releases 2026 AI and ML trends adoption graph
InfoQ released its annual AI, ML, and Data Engineering trends report covering developments through 2026. The report includes an adoption graph tracking technologies from innovators to early majority. New topics added include Agent-to-Agent Protocols like MCP and A2A, and inference engines such as vLLM, SGLang, and llm-d. Inference is expected to account for about two-thirds of AI compute spending by year end.
Kolon Benit trains partners on AI infrastructure
Kolon Benit held a Partner Academy from Oct. 27 to 30 at its AI Solution Center in Gasan-dong, Seoul. The four-day program was organized with Dell Technologies Korea and sponsored by Nutanix. Partner engineers received training on servers, storage, virtualization, and infrastructure management. Topics included Dell PowerStore Elite, PowerEdge 18G servers, Nutanix Cloud Platform integration, and hands-on exercises with tools like Nutanix Move and OpenManage Enterprise.
Human workers face burnout on AI cyber frontlines
A new report finds that AI-related cyberattacks have increased by 300% in the past year. Most attacks are launched by humans using AI tools. The average worker now spends over 50% of their time on AI-related tasks. Many of these workers are experiencing burnout due to high stakes and intense pressure. The report calls for greater investment in cybersecurity and better tools to support human workers.
Semiconductor reliability becomes key for large AI clusters
The engineering behind 10,000-GPU AI clusters is making semiconductor reliability a critical battleground. Companies like Baidu AI Cloud are building large-scale computing infrastructure that depends on PCBs, advanced packaging, thermal management, and power delivery. Investors and semiconductor firms have traditionally focused on NVIDIA GPUs and high-bandwidth memory. Now the focus is shifting to the hidden engineering challenges that keep massive AI systems running reliably.
AI Alarm Monitoring Needs Careful Review and Human Verification
AI coding tools are spreading fast into alarm monitoring software, but industry standards have not kept up. AI can produce code that looks correct while quietly failing on rare signals, which is dangerous in a field where missing events can be life-threatening. Legacy knowledge about panel quirks is often missing from AI training data. The article urges vendors to treat AI output like an unverified alarm signal and have humans confirm it before it goes live.
SecurityHQ Study Finds AI Agent Accountability Gap in Enterprises
SecurityHQ and Datadog surveyed 204 cybersecurity leaders about AI agent governance. While 76% said they have formal accountability rules, 45% reported incidents where responsibility for AI agent actions was unclear. Only 49% could fully reconstruct what an AI agent did in the past 30 days, and just 17% could do so within minutes. Nearly half of leaders suspect unapproved AI agents are being used inside their organizations.
AI Is Causing Stress and Uncertainty for Gen Z Australians
A national survey found that 69% of young Australians aged 18 to 24 feel guilty when relaxing, and AI concerns rose to the fourth biggest worry this year. Cost of living remained the top concern for the third year in a row. About one in four Gen Zers could not go more than three hours without their phone. While some use AI for mental health advice, experts warn that not all AI tools are reliable or trustworthy.
Nvidia Launches RTX Spark Laptops With On-Device AI
Nvidia started selling RTX Spark laptops through partners like ASUS, Dell, HP, Lenovo, Microsoft and MSI. The laptops combine Blackwell RTX graphics and Grace CPU, offering up to 128GB of unified memory and 1 petaflop of AI compute. Starting prices are USD $2,399, placing them in the premium segment. Nvidia held launch events in six cities and emphasized running AI applications locally instead of relying only on cloud services.
Industry Leaders Debate Whether AI Is Reducing Marketers' Importance
At a Chicago event, Sam Wong of PepsiCo and Volles of Jelmar discussed AI's role in marketing. Wong said AI evaluates millions of data points daily and 85% of its recommendations at PepsiCo were accepted. Volles argued humans are still needed for decisions affecting consumer trust. Both agreed that the future will require marketers who can think critically and challenge AI rather than blindly accept its outputs.
AI semiconductor trends shift focus from chips to full systems
Industry experts say AI is changing how chips are designed and built. Instead of focusing on single chips, companies now treat entire packages, racks, and supply chains as one system. Connectivity is now a bigger limit on AI performance than raw computing power. Advanced packaging and AI-driven yield learning have become critical parts of early design. AI agents are also shortening chip development cycles to six to nine months.
Sources
- AI, ML, and Data Engineering InfoQ Trends Report 2026
- InfoQ AI, ML and Data Engineering Trends Report
- Kolon Benit Holds AI Infrastructure Training for Partners
- The humans burning out on AI's cyber frontlines
- The Hidden Engineering Behind 10,000-GPU AI Clusters: Why Semiconductor Reliability Is Becoming the Next Battleground
- Artificial Intelligence-Assisted Alarm Monitoring Development, Part 2
- SecurityHQ Research Reveals AI Agent Accountability Gap Across Enterprise Security Operations
- ‘It’s just a computer, right?’ How AI is tearing Gen Z in two
- Nvidia launches RTX Spark laptops with local AI push
- Mark Anthony, Jelmar Execs Debate on Whether AI Is Making Marketers Less Important
- AI semiconductor industry trends: from chips to systems
Comments
Please log in to post a comment.