OTTAWA — Next year will probably come too soon for artificial intelligence to make its big breakthrough in the business world, OpenText CEO Mark Barrenechea acknowledges. But he’s sure the technology will eventually be used to solve big customer problems.

The Waterloo, Ont.-based firm has gradually shifted from selling one-off software licences to offering recurring cloud contracts. Large clients use OpenText’s technology to manage, analyze and automate their documents and other information, as well as for cybersecurity and IT functions. The company has expanded mostly by acquisition, including a US$6.2-billion deal in August 2022 for British firm Micro Focus.

OpenText reported US$5.77 billion in revenue in its fiscal year ending in June, with US$1.82 billion of that from its cloud business. But while the division continued to grow through the fourth quarter, overall sales dropped after factoring out the sale of the firm’s mainframe-servicing unit. And OpenText’s share price fell sharply after it signalled it would make fewer deals in favour of stock buybacks, as it focuses on increasing cloud revenue organically.  

For now, AI remains a very small part of the firm’s business. But long term, Barrenechea says there’s significant growth potential for OpenText in new applications. The company is building  its Aviator portfolio of generative tools and other AI-equipped products on top of the large language models and computing services of other tech firms. While any other company can employ the same raw materials, Barrenechea says OpenText can still compete in AI because of its history of automating important processes within its clients’ operations, and managing security and access for that information.

In an interview with The Logic earlier this month, the long-serving CEO discussed OpenText’s plans to build out its AI business, the practical challenges slowing the technology’s spread and how the firm is using it internally. 

This transcript has been edited for clarity and brevity. 

What specifically are your customers using AI for?

We’re at the apps layer. We have these decade-built, very pristine data sets that are continuously getting stronger from automation. We’re going to embed AI into all our major functions. Our customers are focused on very big problem sets. 

We’re working with a major insurance company to analyze risk. They think they have an approach in auto, reinsurance and a couple other markets to break through. Those who can evaluate risk become leaders in underwriting. 

We’re working with a major apparel company in supply-chain optimization around shipping, shrink, and return-to-vendor. Five per cent optimization could save half a billion a year through AI. 

We’re working with a very large industrial manufacturer that has over 50,000 field service workers, to solve worker safety in physical infrastructure, from power plants to power lines to trucks and construction.

I’m hearing from a lot of executives that last year was really one of AI pilots, proofs of concept and testing. Was that also OpenText’s experience with customers?

For sure. My best way to describe exactly what you said—and we experienced it—is Amara’s law: You tend to overestimate the effect of a technology in the short term, and underestimate [it] in the long term. 

This is one of the most consequential technologies of our lives. We see the value we can provide. But the cost remains high. Like any capital infrastructure project, you want to make sure you get the right vendor, outcome, and cost. 

There were pilots, experimentation. What we’re seeing coming into our [new] fiscal year are big projects. Customers are focused on solving big problems, and they’re not going to go public until they’ve solved it, because this is going to impact their direction, their earnings, their competitiveness. 

I’ve also heard that a relatively small share of these pilots are actually converting to real-world, day-to-day use. Is this turning into real revenue for you?

It will translate to real revenue, but it’s going to take time. I don’t think fiscal ’25 is the breakthrough year. We’re winning bookings [and] revenue, no doubt. But we’re also not over-investing. 

What’s the breakthrough going to take? It’s going to take bringing costs down and tuning the automation in the AI so any capital investment is sustainable over a decade or so. I expect to announce some big productions in fiscal ’25.

So 2025 might not necessarily be the breakthrough year—

It’ll be a year of wins, a year of progress. I like the railway mania analogy. The U.K. government laid over 2,000 miles of track in 1845. It was up to seven per cent of their GDP and almost 50 per cent of their infrastructure investment in that year. It only produced £30 million in revenue. It took 10, 15 years for that infrastructure to get to £2 billion in revenue. 

A lot of mania, a lot of entrepreneurialism, a lot of crazy ideas—open train cars pulled by horses! Not everyone made it to the other end. But they got the infrastructure right in the peak of spending, and created their industrial power [and] autonomy. 

We’re a believer. There’s a breakthrough moment coming on refinement of automation, data [and] models, and [on] the cost coming down.

One of the examples you were giving me was supply chain optimization, where a five per cent optimization could save half a billion a year. Can the technology provide that?

The short answer is: It’s getting close. It’s a combination of: How good is your automation? How good is your data set? How good are your learnings and your ability to interpret the results? 

Let me establish the base here. Last year, OpenText’s cloud revenue was US$1.82 billion. What share of that came from AI?

Very small. About five per cent of our revenues are related to AI today. 

We’ve seen a bit of an AI stock sell-off recently, where the larger tech names haven’t been able to deliver quite the revenue growth that the market expects. Is telling any kind of AI story now a stock-price risk because of the expectations that it generates?

I don’t believe so. I’m not in consumer AI, I’m in business AI. It’s in every RFP and every discussion. If it’s 1845 and 2,000 miles of tracks just got laid, that’s okay. 

We’re going to keep talking about our business clouds, business AI, and business technology like security and IoT. I won’t shy away from it. If you think it’s part of your competitive advantage, you need to talk about it. And we are.

What are you doing internally with AI? 

We have an internal platform called Olie, which has been our knowledge base for the last decade—all our support, customer interactions, contracts, product stories and narratives. We’re a knowledge-management company, so we use our own software. We’ve applied Aviator to it. We’re running on a GPU farm. 

We have applied it [to] tech support—we are elevating how we interact with customers. We’re now responding to RFPs in pre-sales with AI; there’s still a human review. We’re also now using it to deliver professional service code—configurations, customizations. 

We have a second project underway called Platform Athena, [in which] we’ve actually developed our own language model for engineering. We’re going to have our first AI-generated software—screens, APIs, translation, documentation—generated by the end of this year.