By Forbes Staff,Megan Poinski

Copyright forbes

AI brings many possibilities to an enterprise. It can streamline workflows, enhance productivity, improve efficiency, automate tedious work, facilitate deep research and perform simple tasks. But it is also disrupting IT governance and oversight. A new report from governance platform OneTrust quantifies how AI has transformed how companies work through data governance and risk issues. Companies report spending an average of 37% more time on AI governance now than 12 months ago.

But it isn’t just that AI requires more time spent on governance. It’s completely changing the governance picture. Three-quarters of business leaders said that AI exposes the limitations of their legacy governance processes, with almost the same number reporting critical gaps in visibility, collaboration and policy enforcement. Nearly all respondents—82%—say AI risks are accelerating timelines to modernize their governance strategies, and more than half expect their governance budgets to go up more than 20% next year.

An overwhelming majority of companies told OneTrust that they’re underestimating at least one risk that AI brings, with the largest contingent—44%—concerned about cybersecurity vulnerabilities. However, more than a third are also concerned about third-party uses of AI, gaps in data governance and issues with AI agents. And working through governance to address these issues can be difficult. OneStream identified several issues with implementing necessary changes, with the two most common being a lack of integration between governance tools and AI and limited budgets.

As the wide variety of risks associated with AI demonstrates, along with the rapid adoption and evolution of new tools, governance challenges should be addressed sooner rather than later. The study suggests tackling the root causes of the problems—often outdated infrastructure and processes—and recreating something more streamlined and appropriate for the technology. Going forward, governance should be an integral part of every new AI tool adopted, with an eye on preventing major governance issues in the future.

Enterprise platforms are bringing in more AI tools to both enhance their offerings and provide customers opportunities to streamline their work. Enterprise planning system Asana recently named Dan Rogers its new CEO, and the platform is adding more AI agents to its workflow. In one of his first interviews as CEO, Rogers spoke with me about the challenges with AI agents, and how enterprises can improve their work. An excerpt from our conversation is later in this newsletter.

This is the published version of Forbes’ CIO newsletter, which offers the latest news for chief innovation officers and other technology-focused leaders. Click here to get it delivered to your inbox every Thursday.

NOTABLE NEWS

Nvidia CEO Jensen Huang at the Consumer Electronics Show in January.
PATRICK T. FALLON/AFP via Getty Images

Last week, Nvidia made headlines for its $5 billion investment in Intel. The world’s most valuable company hasn’t put its checkbook away just yet. Nvidia announced plans on Monday to invest up to $100 billion in OpenAI, creating a partnership that focuses on developing AI data centers and superintelligence. Through this partnership, at least 10 gigawatts of Nvidia systems will be deployed for OpenAI to train and run its next-generation models, working toward building a more powerful compute infrastructure. The first gigawatt of Nvidia tech through this agreement is slated to go online in the second half of 2026.

Nvidia also put $500 million toward Nscale, a relatively new player that runs GPU data centers, writes Forbes’ Iain Martin. The startup, which got its start under a crypto mining company that was spun out in 2023, pivoted to powering AI startups. As of last month, Nscale CEO Josh Payne told Martin the company has 40 megawatts of AI capacity online. However, it has several large contracts—and lots of promise. At an event with the U.K. Prime Minister Keir Starmer during President Donald Trump’s visit to the U.K. last week, Nvidia CEO Jensen Huang said, “Three months ago, when I met Josh, they were zero billion dollars [revenue]. Today, Josh is zero billion dollars [revenue], but with potential.”

Huang said Nvidia has helped get Nscale set up for success. In July, Nscale landed a deal with OpenAI to build its first “Stargate” AI data center in Norway’s Arctic Circle. Last week, Nscale announced a similar deal for a “Stargate” data center in the U.K., as well as an agreement with Microsoft to build the U.K.’s largest AI supercomputer.

Nvidia has backed several smaller companies in the AI space, both ones providing infrastructure support and others making use of the technology. According to Crunchbase, Nvidia has invested in several AI companies just this month, including Figure, Mistral AI, ElevenLabs and Scintil Photonics.

POLICY + REGULATIONS

President Donald Trump signs an executive order introducing a $100,000 fee for H-1B visas in the Oval Office last week.
Andrew Harnik/Getty Images

The U.S. tech industry employs workers from many different countries, historically leveraging its dominant global position to attract the brightest and most innovative people from around the world. But a recent proclamation from President Donald Trump may change that. Last week, he hiked the fee for H-1B visas, commonly used by companies to allow specialized foreign workers with college degrees to stay in the U.S., to $100,000—up from $780 for employers who petition to receive the visas for their employees, or $215 to be selected through a lottery. There’s an annual cap of about 85,000 H-1B visas for private companies, and many tech companies rely on the program. The new fees only apply to new visas and workers, not renewals or people already in the U.S.

The announcement caught the tech industry by surprise, writes Forbes senior contributor Emil Sayegh, and could have negative impacts on U.S.-based companies. With the cost of bringing an employee to the U.S. outpacing some entry-level salaries, startups may find themselves limited in hiring. Forbes’ Richard Nieva and Rashi Shrivistava write that some European tech startups see the new visa fee almost as a gift from Trump. But while some global companies may build their tech talent in countries that are friendlier to international workers, there are tech jobs that shouldn’t be offshored, such as cybersecurity for critical U.S. companies.

The fee hike fits with the Trump Administration’s pushback on immigration. In August, Commerce Secretary Howard Lutnick said the program is a “scam that lets foreign workers fill American job opportunities,” and Vice President JD Vance has accused tech companies of laying off American workers to apply for H-1B visas.

According to federal data, Amazon employed the most workers (10,044) using H-1B visas as of June 30, followed by Tata Consultancy Services (5,505), Microsoft (5,189), Meta (5,123), Apple (4,202) and Google (4,181). The new fee disproportionately affects talent from India—about 73% of H-1B applicants approved in FY 2023 were born in India, and the subcontinent has been the largest beneficiary of the visas every year since 2010.

CIO STRATEGY

Another new study on AI in the enterprise has come to the same conclusion: A siloed workplace hinders AI success. Forbes senior contributor Joe McKendrick writes about a study from FICO, which found that only 5% of financial enterprises have AI investments aligned with their business goals. The numbers echo an MIT study released last month that indicates just 5% of generative AI projects are showing tangible success. FICO’s report showed that most companies are optimistic about the possibility of AI to bring positive changes in the next five years. But the problem lies in outdated org charts when it comes to AI projects.

“The problem isn’t ambition,” the FICO study states. “Most organizations now have dedicated teams focused on data science and AI development. But too often, these teams operate in silos, cut off from business leaders, end-users, and risk stakeholders.”

Nearly three-quarters of respondents said that insufficient collaboration between business and IT stakeholders is the biggest challenge to organizational alignment. This dwarfs all other issues, including resistance to change and lack of strategy. Working together and real collaboration across all departments represents the best solution.

BITS + BYTES

Asana’s New CEO Talks About Improving AI Agents

Asana CEO Dan Rogers

At the end of July, Dan Rogers succeeded Asana cofounder Dustin Moskovitz as the enterprise planning platform’s CEO. Rogers is a tech veteran who most recently served as CEO of the software feature management platform LaunchDarkly, and also has stints at cloud data management company Rubrik, business workflow platform ServiceNow, Salesforce, and AWS.

I spoke with Rogers in one of his first interviews as Asana’s CEO about the company’s push to use AI agents as a path to productivity and the stumbling blocks they face. According to statistics shared today at Asana’s Work Innovation Summit, 63% of employees believe that AI agents are unreliable, and there is no clear rule for who is accountable for AI agent failure. This interview has been edited for length, clarity and continuity.

Why is there so much discontent with AI agents right now? Why aren’t they doing what employees want?

Rogers: There’s two groups of discontent. The first is [concerned] about: Are they achieving the productivity gains that we hoped that they would achieve? The second is: If they are going to achieve those productivity gains, what does that mean for our jobs and work as we know it?

I’ll address both of those. First is about productivity itself. You can’t expect all of the Jeopardy winners [coming together to offer training to people at your workplace] to be immediately productive because they don’t have the [context] to run upon. When Asana built our work graph, we built this data model that had a very specific description of how these things need to work together. This concept that we want to understand the relationships between people and between their goals, the why of what they’re doing, and the context of their work.

Without that true context interacting with people, agents don’t really have anything to work upon. They could only work within a very narrow domain. They would require a lot of oversight and guidance. So to achieve productivity, you need something like a work graph, connecting to the relevance of everything that’s gone before in a project and the systems that it takes to complete. That’s the productivity unlock piece.

Then there are some fears in the broader community around: What does this mean for work as we know it? What does this mean for our jobs?

I have a very positive disposition. My son will graduate and go into the workforce at some point in the near term. What do I think the future is for him? I see it as one of creation. I see it as one of building.

With these new, powerful, tools and techniques, we will be able to solve the long tail of humanity’s problems. There will be new things that we will uncover and discover. This is a productivity boost, as opposed to something that we should fear. It’s something that I believe that we should embrace. And when I see the next generation playing with these tools, I see nothing but this real spur and flourishing of creativity and creation.

How do you get there? What are some of the components needed to get the productivity on track, to reach the creativity potential of AI agents?

The answer is not autonomous agents. The answer is teammates. The answer is human AI collaboration. And for that to happen, you need to have context.

If you think about the foundational models, even for our personal lives and our personal productivity, they need a lot of context. The more context they have, the better they are. This is true within an enterprise to drive enterprise productivity. These agents, this kind of AI helpers, need to understand the context of what we’re trying to do. They need to understand the alignment to goals, what’s been done in the past, who’s done it. So context and productivity are very closely linked.

We also believe that because it’s a period of human-AI collaboration, you are going to want to have checkpoints. Checkpoints happen early, so when you are creating the prompts for the agents, having early tests that the responses are actually coming back in the right way, the checkpoints are also initiating those next phases of work and understanding what the empowerment levels are for the agent to complete that work. What kind of decisions do you want to empower those agents to make, and where do you want to have a human in the loop?

Finally, this idea of controls is very important. Just because you are working with a vendor that’s AI-forward, it doesn’t mean that you can cast aside all of the existing governance controls that you already have. You already have data policies, you already have billing policies. The agent needs to adhere to those. They just can’t run on some other schema that is actually making your data publicly available, or doesn’t have the same approvals as the rest of your workers.

These kinds of things—better context, alignment to goals, checkpoints and controls—should be apparent to enterprise leaders bringing in AI agents. From what you see, why are they being pushed to the side? Are companies more looking to bring in AI for the sake of AI?

I’d describe it through some of the framing of how new technologies get introduced into the world. I had a front-row seat on a few of these transformations: [For] on-premises infrastructure to cloud infrastructure, going from manual workflows to digitized automated workflows and of the transition from human-generated code to AI-augmented code generation for developments. I’ve seen these transformations happen a few times already.

What tends to happen is people get very excited, and I think it tends to be a lot of hope and a lot of promise. They come through a trough of disillusionment. Coming out of that trough of disillusionment, real productivity does actually happen when you find those specific use cases, when you go back to some of the business principles for how you improve those workflows.

A tool maketh not the answer, right? The answer is a lot of the change management that’s going to be required to get the productivity that you want out of some of these new techniques. But nonetheless, with the cloud transformation, with a digital transformation, these things do happen. They just don’t necessarily happen in that first wave.

This is why I’m excited to join Asana. I’m excited to build this with our customers: What this era of human-AI collaboration is going to look like, how we get to an agent platform that drives outcomes with the right context, controls, checkpoints.

COMINGS + GOINGS

Industrial products supplier Cooper Standard promoted Soma Venkat to senior vice president and chief information technology and AI officer, effective September 15. Venkat had been working as senior vice president and chief information technology officer since 2021.

Defense technology firm Kratos Defense & Security Solutions elevated Brian Shepard to chief information officer. Shepard was most recently vice president of information technology, and worked with Haverstick Consulting prior to that, which was acquired by Kratos in 2006.

AI automation and security provider Infoblox appointed Justin Kappers as chief information officer. Kappers joins the company after working at NetApp, where he served as vice president of enterprise applications.

STRATEGIES + ADVICE

AI coding makes it easier to write new software and applications, as well as release new versions of existing things faster. However, it increases your cybersecurity risk—especially as your code volume grows. Here are some ways to keep track of your code volume and be on top of security issues.

We’re still a few months away from 2026, but the newest trends around AI are already beginning to surface. Here’s a look at the eight AI developments most likely to shape the business world next year, and what you can do to get ready for them.

Several powerful entities—including the U.S. government and Nvidia—have announced recent investments in Intel to help the chipmaker reinforce its business. According to a Bloomberg report, the chipmaker is reportedly seeking additional powerful investors. Which company did Intel recently talk to?

See if you got the answer right here.

Got a tip? Share confidential information with Forbes.

Editorial StandardsReprints & Permissions