Q2 2026, year over year

It’s hard to look away from AI, since it’s pretty much everywhere. If we manage to do that for a moment, the pace of change in world events, politics, and the environment has been equally intense. After pausing for a moment to wring our hands, let’s get some facts about the direction and pace of change this past quarter vs. a year ago.

Looking back

Q2 2025 was the height of experimentation. About 88% of enterprises used AI somewhere in the business, but only 22% had scaled it across the organization. Budgets reflected this early stage, with most spending focused on model testing, cloud computing, and small automation pilots.

By Q2 2026, 40% of enterprises reported organization-wide scaling, and 87% of large companies had at least one production deployment. Average enterprise AI budgets rose more than 60% year over year, and global AI spending projections for 2026 exceeded $2.5 trillion, driven by infrastructure, orchestration layers, and integration tooling.

Data centers and resource use

Concern in Q2 2025 was warming, as were air and sea temperatures. Data center electricity consumption rose approximately 8% as enterprises put AI to work. At those workplaces, human concerns rose even faster. Water usage for cooling increased less dramatically, by 6-8%, wherever evaporative systems were in use. That reflected a growing gap between AI potential and physical capacity.

Resource demand accelerated by Q2 2026. In several regions, electric utilities attributed 12-18% additional consumption to data centers. Water usage bumped up by 10-15% as facilities expanded their cooling capacity.

The use we see

In Q2 a year ago, 65% of enterprises had assigned at least one business function to generative AI. The assignments were fragmented into many types, in many places. Multiagent systems had just hatched, with adoption around 26%.

By Q2 2026, generative AI had become standard across many workflows, and multiagent systems were moving into everyday operations. Industries under automation pressure saw year-over-year adoption increases of 14-17 percentage points. Workforce expectations shifted as AI fluency became a baseline requirement.

How that adds up

Q2 2025 optimism met Q2 2026 realism. Only 29% of companies reported strong ROI from generative AI, and 34% of projects never reached production. Integration complexity, workflow misalignment, and data quality issues were deemed responsible for all that.

In this year’s second quarter, companies that invested early in AI infrastructure, governance, and workforce readiness seem well positioned to benefit in late 2026. For readers who want to explore the numbers further, two helpful sources are AIStackHub’s quarterly adoption surveys (aistackhub.com) and the Global Enterprise AI Scaling Report (enterpriseaiscaling.org).

A Fable unfolds

Have you met Claude? I recommend Claude AI (claude.com) to writers who are protective of their work, whether they are revising a novel in progress or developing proprietary content for their workplace. Claude offers the option to delete a session, giving writers control over what they and the AI explore together. The ability to remove a conversation limits the risks of experimenting, drafting, and thinking aloud. It helps many writers feel more comfortable using AI as part of a creative or proprietary process.

Now Claude Fable has emerged. Fable is Anthropic’s general‑use version of its latest frontier model. It shares the same underlying architecture as Mythos 5, the high‑power cybersecurity variant, but is wrapped in safeguards that allow broad deployment. In short, Mythos is restricted, and Fable is the public‑safe version of the same intelligence.

After arriving in early June, Fable disappeared within its first week when the U.S. government applied export‑control rules requiring user nationality to be verified in real time. That put it into blackout from June 12 through the end of the month, when export controls were lifted. Fable redeployed on July first with new safeguards and new usage rules.

Among the rules, Fable is not part of the free tier. It resides within Claude’s paid plan, allowing time‑limited access for about twenty dollars per month. Before reaching for their wallet, writers should remember that Claude’s free plan still includes the session‑deletion feature that many consider essential. When a project calls for long‑context reasoning, multi‑step analysis, complex coding, or a blend of structure, creativity, and logic, it may be time to bring out the plastic. Anthropic has woven a compelling suite of high‑end capabilities into Claude Fable.

Today’s legislation about tomorrow’s data centers

We’re seeing that AI is outgrowing its foster facilities. Several hundred data centers across the United States are now hyperscale, built for cloud platforms and AI training. Some of them use as much electricity as a small city. Some use millions of gallons of water per day just to stay cool. Energy and water bills in neighboring communities reflect the impact, and the summer heat is not helping.

That’s the backdrop for today’s legislation. New York lawmakers may pause development of hyperscale data centers for one year. The timeout being considered in Oklahoma would stretch to 2029. Vermont legislation might delay development until 2030.

Lawmakers in more than two dozen states are debating the impact and cost responsibilities of grid upgrades. Among the federal bills crawling through committees, one asks whether data centers should be required to generate their own electricity.

Online resources are available for keeping an eye on this. MultiState.us tracks data‑center energy rules and proposed moratoriums. The Data Center AI Laws by State tracker at ailawsbystate.com maps where pauses, new reporting requirements, or environmental reviews are underway.

Join a virtual wagon train, headed west

This summer is the season for exploring the frontier of workplace technology. If record-breaking heat or the discomforts of air travel concern you, three major conferences will stream live to provide a virtual vision into AI-empowered collaboration and immersive tools.

Microsoft Inspire (July 14–18) may originate from Seattle, but it arrives in a fully virtual format that highlights a growing set of AI‑driven tools for writing, analysis, and team collaboration. If history repeats itself, we are likely to see demonstrations of devices designed for hybrid work, including laptops, headsets, and accessories that pair well with Microsoft 365. Details and registration are available at events.microsoft.com/inspire.

The Ai4 2026 wagon train rolls into Las Vegas in early August and offers a virtual livestream pass. It’s designed for those of us who want insight into the coming attractions and concerns of enterprise AI. Sessions will focus on automation, analytics, and responsible AI, and the program provides a look ahead at ways that these systems may shape daily work. More information is available at ai4.io.

Another Microsoft offering, TechCon 365, arrives in Seattle in late August. It also offers a viable virtual option that brings together Microsoft tools such as Copilot, Teams, SharePoint, and Power Platform. Virtual labs and demos will feature tools such as the Meta Quest 3 headset for immersive meetings and the Logitech MX Keys S keyboard for AI‑assisted writing. Registration details are posted at techcon365.com.

Though originating from the west, the conference trains are headed toward the future of work. You heard it here first: the next frontier is not a place but a platform.

The new shape of the workplace

We’re likely to be at our desks on Monday mornings, whether at a workstation or in a home office. Whether those around us are humans or felines, our work has evolved.

We can learn about the latest in workflow design at SAP’s AI Adoption & Learning Live (July 6–9). The event is virtual and free. It’s also modular, with drop‑in sessions that offer perspectives on AI in the SAP ecosystem. For those who dwell outside that ecosystem, other tracks provide a chance to see how teams can adapt to working with AI without being overwhelmed. More information is posted at learning.sap.com/artificial-intelligence/adoption-learning-week.

A second option, the AI Communications Virtual Conference, will be held on July 22 and hosted by the long‑established publisher and training organization, Ragan Communications. The conference (priced at $1,699) explores how AI is reshaping internal messaging, content workflows, and organizational communication. Further details of this and similar events are listed at ragan.com/events.

Looking back

We see remarkable evolution in just a few years. About 28 percent of workplaces used AI tools in 2024. By 2025, that rose to roughly 41 percent, and by the time we’re spiking the eggnog at the dawn of another year, more than half of U.S. teams may rely on AI in daily workflow. Events like those offered by SAP and Ragan can help leaders keep pace as the pace of change in work environments and tasks accelerates further into this century.

Writing for the machines

Their listening devices are everywhere. Your desktop and your phone hear you without a wake word. Words that are typed or spoken travel through silicon before any human hears or reads them. Our smart helpers parse and reshape the systems we write through.

The Center for Plain Language (centerforplainlanguage.org) offers an on‑demand session, Practical AI for Plain Language Professionals. It’s a $65 walkthrough of promptcraft and risk‑spotting wherever machines are your first audience. That’s everywhere, by the way.

Since they’re listening, I should also mention Microsoft’s free Copilot for Writing Workshop. It’s a short tour of how AI can support workplace writing. I won’t include the web address because you already know where to find them, just as they know where to find you.

World Productivity Day lands on June 20, so this is a good week to reconsider how we collaborate with the invisible infrastructure that speeds up our work, slows us down with cat pictures, and underscores our words with squiggles along the way.

A virtual visit to AI Con USA 2026

Why worry about airfare or $5-per-gallon cross‑country to Seattle? AI Con USA 2026 (June 7–12) offers a full virtual track. The free virtual pass includes the keynotes, select concurrent sessions, a virtual expo, plus access to session materials for most talks.

This year’s theme of real‑world AI focuses will take deep dives into Large Language Models, Retrieval‑Augmented Generation, and the emerging wave of agentic AI workflows.

On-site attendees will nosh and network with hundreds of tech and engineering leaders, researchers, and enterprise professionals, while thousands more of us participate online. Check your daytimer, and if you are following the shift from prompting to building AI systems, here’s one to watch. Details and registration are available at aiconusa.techwell.com.

Origins

AI Con USA has been steadily growing into one of the more practical, engineering‑focused AI gatherings in the country. The 2024 conference centered on early, hands‑on adoption of Large Language Models, with sessions on testing AI systems, building first‑generation RAG pipelines, and helping software teams integrate AI into existing workflows.

The 2025 event expanded on that, adding more case studies, more mature RAG architectures, and the first wave of agentic workflow sessions. TechWell reported record virtual participation as the conference leaned into the idea of real‑world AI as something teams build, refine, and maintain.

In those two years, AI use has emerged from experimentation to large-scale deployment. Having seen technology evolve so quickly, we expect that AI Con USA 2026 will be especially worth watching.

Workplace writers can have an agent

As of this week, it seems that AI agents have arrived and are ready for duty. If you would like to learn more about them, today would be a good day to secure a free virtual pass to AI Con USA (June 7 through 12). Two streamed keynotes focus directly on what agents can do.

AI agents run on the most popular free browser-based and desktop AI apps. They are available by opening one of these apps, creating a new workspace, assigning a task to the agent, and granting it permission to act. The app handles the planning, the steps, and the execution.

As you learn the agent-recruiting skills of framing a task and setting boundaries, you might want to attend Will the Real Autonomous Agent Please Stand Up and Architecting the Digital Work System: The Shift to Multi-Agent Enterprise Workflows, both live-streamed from AI Con USA.

AI agents are a recent development. They stayed in the background until late 2023 and early 2024, when researchers demonstrated their planning, revision, and corrective capabilities. AI Con USA will explore those topics. Your free pass to the virtual sessions awaits at aiconusa.techwell.com.

Season’s findings

 May is when Microsoft publishes its Work Trend Index (WTI) annual report. It’s now just a Google (or Edge) away.

Amid the website’s bells and whistles, you’ll find plenty of cyber-fueled optimism supported by findings from surveys of “20,000 workers using AI across 10 countries.” The WTI report has a message for leaders: “organizational factors… account for twice the reported AI impact of individual effort alone.” The organizational factors cited are culture, manager support, and talent practices.

At a high level, Microsoft’s data splits AI use into these categories: 49% - Analyzing/reasoning/deciding, 19% - Interacting with others, 17% - Producing work, 15% - Information gathering. Removing the gerunds from those categories, you might say that your AI Assistant spends almost half of its time as a Senior Manager or Executive, with the other half of its 24/7/365 workday fairly evenly divided among communication, content production, and research tasks.

Echoes from recent years

That AI Assistant has only recently been onboarded. In May, 2024, Google expanded its workplace presence with generative AI features. One year later, Microsoft’s WTI reported a widening gap between AI adoption and organizational readiness. In June that year, Newsweek presented the AI Impact Awards, recognizing large organizations that had demonstrated measurable productivity gains through AI‑assisted work.

Compared with last year, the recent WTI report shows that onboarding can have an impact, driven by improvements in organizational factors that have led to better workplace use of AI.

Log on and choose an option

 Two events on Thursday, May 14, offer sharply different ways to think about how AI is reshaping work.

At MIT Sloan Management Review’s Work/26: Leading Through, scholars from MIT, Harvard Business School, and Wharton join executives from Fidelity, Mercer, and S&P Global to translate their research and experience into guidance leaders can apply immediately.
(Navigator Pass – $99, 10 a.m.–2:30 p.m. EDT, events.sloanreview.mit.edu)

For $99 less than that, the AI Skills Conference begins at 11 a.m. ET and runs for more than five hours. Over 5,000 professionals across digital, product, data, and marketing teams will gather to share practical AI workflows and real business use cases.
(Session details and a link to free registration are at hackernoon.com)

The event times overlap, so if leadership insight is your focus, the MIT Sloan event might be a good $99 learning opportunity, but to get down into the weeds of implementation, a five-hour investment in HackerNoon’s AI Skills Conference is certain to return value. Either way, Thursday offers to bring a preview of tomorrow’s AI workplace to a browser near you.

It seems we’re hearing echoes

Four years ago this month, the First Workshop on Intelligent and Interactive Writing Assistants brought together researchers who presented papers on how humans guide machines to produce good writing and on how cognitive writing models might shape future tool design. Even before ChatGPT went mainstream, the potential vs. peril of writer collaboration with AI was being studied and vigorously discussed.

An online learner’s odyssey

Even if airfares climb to new altitudes, the world of AI at work can still be explored without leaving your desk. Three conferences next month originate thousands of miles apart, yet each is fully accessible from any home or corporate campus.

Informa Tech’s AI Summit Series streams from London on June 10–11. A visit to theaisummit.com offers a look at how companies frame their AI strategies, define responsible use, and shape messages that bridge technical and non‑technical audiences.

A few days later, you can shift your attention to San Francisco for Databricks’ Data + AI Summit (databricks.com). The free virtual pass covers June 15–18 and focuses on turning complex data and AI workflows into clear, actionable insights.

By month’s end, the AI Engineer World’s Fair brings you to Austin, Texas, on June 29–July 2. Its online hands‑on demos and workflow sessions will give writers a practical sense of how today’s tools actually function.

The travel cost for the entire itinerary is zero, though no one will stop by your desk with a bag of chips or a refreshing beverage.

A look at AI in everyday work

WorkCongress 2026: Future of Work & AI will assemble an online multitude this Friday to consider how AI is reshaping everyday work. The virtual convocation will bring together experts, practitioners, and leaders to explore AI applications for productivity, collaboration, and decision-making.

The 100%-free sessions and demos will be hosted at WorkCongress.com. Pre-registration, sans credit card, is available there today. Return on Friday to enjoy a look at how AI can support teams, streamline workflows, and open new possibilities at the workplace.

Last year’s perspective

One year ago, Meta’s LlamaCon marked a moment when open‑source AI felt as if it had stepped fully into the center of workplace innovation. The conversations that week revolved around Llama 3 and the steady improvements shaping its model family. Performance was rising, context windows were stretching, and open models were beginning to move from the waters of research onto the dry land of everyday equipment.

Llama has grown since then. Installable models are available at llama.com, and Meta is preparing the Llama Application Programming Interface, which will run in the cloud rather than on your own hardware. The API is in preview for now, and you can register for access at the same site.

Could it be that AI is settling into the rhythm of everyday work? Last year, the conversation was about the potential of open models. This year, we’re hearing about what they already do.

Webinar brings new tools to the pipeline

If you’re in corporate sales, either customer-facing or in leadership, the Sales Transformation with AI webinar offered free by Microsoft might be worth an hour of your time tomorrow (4/23/2026). You’ll learn how AI might join the sales team to automate routine tasks, enhance customer experiences, and find revenue growth opportunities.

The webinar begins at 9:00 am Eastern time. Register for event id #4071837195 at events.microsoft.com.

Autonomous agents arrive

Three years ago, Stanford University released a Generative Agents research paper, that ignited interest in autonomous workplace assistants. Also that year, surveys from Deloitte and McKinsey found that over 60% of organizations had experimented with AI-driven workflow automation.

By 2025, Boston Consulting Group reported that 78% of organizations were using AI tools directly in their workflows.

Looking into workplace AI…

Tomorrow’s workshop from JourneyTeam gives us a chance to see AI “under the hood.” Their Microsoft Copilot & Agents Interactive Briefing: See AI at Work will demonstrate how Copilot and custom agents run workflows inside Microsoft 365. You can register for that on the journeyteam.com website; choose “Webinars” from the “Resources” menu.

…and back at these milestones

Eight years ago this week, Google brought Smart Compose into Gmail. Since that day, AI has been available to assist with drafting emails. Then, in 2021, OpenAI introduced natural‑language programming and automation, making technical tasks accessible to people well beyond the I.T. department.

Gallup has looked at broader workplace trends and found that employees who use advanced digital tools are significantly more engaged. It seems that AI steadily moves closer to the center of daily work.

What’s taking shape…

Why worry about air travel’s long lines and cramped cabins, when events like HumanX 2026 (4/9/26) and the MIT AI Conference (4/14/26) can be attended online?

Both show potential for views of where workplace AI is headed next. HumanX leans into the ways teams progress from first investment to returned value. MIT has research-grounded guidance on deploying and sustaining AI as a workplace tool.

You’ll find sessions at either event that offer insight into AI governance, metrics, and scalable adoption. If those are among your current interests, more details can be found at the HumanX and the MIT Industrial Liaison Program websites.

…and what shaped it

It was only three years ago that generative AI entered the mainstream of business operations. Interest in agentic AI quickly followed.

Just in case generative and agentic are not yet favored buzzwords in your neighborhood, here’s what Mr. Google says about them:

Generative AI (GenAI) focuses on creating content (text, images, code) based on prompts, acting as a creative engine. Conversely, Agentic AI acts as an autonomous manager that pursues goals, strategizes, and executes multi-step tasks across systems independently.

Translated into human, that means generative AI helps us with words and pictures, while agentic AI takes action.

This week’s learnings…

Taken together, GitHub’s Copilot Innovation Workshop and the Azure Virtual Training Day: Data Fundamentals, both on April 2nd, show ways that AI now reaches into everyday work. The Azure session agenda goes beyond the tech subjects of data and automation to explore daily development workflows. Both online programs are certain to suggest ways that AI (and Microsoft’s Copilot in particular) can save time and improve productivity.

…and last week’s lessons

AI tools for remote work planning and scheduling brought immediate value in 2020 as the COVID-19 pandemic sent workers home. Three years later, organizations were testing everyday use cases for generative AI.

In 2024, time management assistants became mainstream. There’s been a shift from experimentation toward policy-defined AI-use efficiency. Recent Gallup research shows knowledge workers save several hours per week using AI. McKinsey reports that over half of companies now embed generative AI into their core workflows.

This week’s focus…

The AI in the Workplace webinar, offered by QuestCover (Quest Business Services) (HR Webinar) this week, will present recommendations for introducing AI into everyday work. Its human resources perspective will focus on compliance and management responsibilities.

Another event, running March 30–April 1, will take a broader view on AI. It’s a global OECD AI WIPS Conference, just a few clicks away on the OECD Events site. Management practices across industries will be shared in sessions for professionals seeking additional ideas for working most productively with AI.

…and its foundations

The release of OpenAI’s Gym toolkit in early Spring 2017 marked a turning point from experimentation to everyday workplace use. The toolkit standardized “reinforcement learning”, a trial-and-error process for decision making by receiving positive or negative feedback for actions taken. Who among us hasn’t learned in that sometimes-painful way?

Four years later, and at this time of year, Microsoft introduced AI tools through its Azure cloud platform. Among the services to support AI implementation were governance, transparency, and tools to help organizations manage ethical risks, especially in workplace environments with regulated, public-facing applications.