SAS rejects generative AI hype in favor of data fundamentals at Innovate 2025
SAS Innovate 2025 is over and with it, three days of keynotes, hands-on sessions, and roundtable discussions focused primarily on data and AI have come to an end.
Mood on the ground throughout the conference was high, with customers and partners alike paying testament to the analytics veteran’s nuts-and-bolts announcements and realistic outlook on how businesses can benefit from technology such as AI.
Throughout this year’s event, SAS has repeatedly taken the opportunity to pour cold water on generative AI hype, stressing that while the technology is great for certain tasks it isn’t a replacement for proper data analytics, nor ‘traditional AI’ such as machine learning (ML) or neural networks.
At the opening keynote Bryan Harris, CTO at SAS, made the sarcastic claim that generative AI could make enterprise problems “magically disappear”, before reminding attendees that LLMs come with limitations such as hallucinations and a tendency toward bias.
Announcements on domain-specific AI and collaborative agents show SAS is invested in making AI easier to adopt and more effective off-the-shelf for organizations in specific sectors.
The company is also not shying away from the agentic AI trend, with SAS Viya Intelligent Decisioning allowing customers to build and deploy custom AI agents.
But in its send-ups of AI hype and repeated calls for strong AI governance, SAS also spent a good portion of its annual event tackling the assertion that generative AI is the best technology for every problem.
Jim Goodnight, co-founder and CEO at SAS, gave a scathing assessment of generative AI in conversation with ITPro.
“There’s a lot of money being spent on generative AI and there’s a lot of money being wasted,” he said. “The results end up pretty negative because so many companies are just trying to take their current staff and say, ‘Okay, go do this generative AI stuff and they do that and they spend a lot of money.
This wasn’t limited to Goodnight alone. Gavin Day, the newly-appointed COO at SAS, told ITPro some enterprises could be facing a “summer of disappointment” as far as AI ROI is concerned.
Of course, SAS is itself using generative AI, with Goodnight quick to point out the value for customers in using LLMs to turn natural language prompts into code outputs written in the custom SAS programming language. He described the AI-generated code as “actually pretty good”, comparing it to the work of a novice.
“It’s like your first program that you ever wrote, this is sort of the level of the code that’s generated, so it’s not super professional but it’s maybe 80-90% correct.”
All of this adds up to a defiant approach, in which SAS provides customers with tools to prepare for, build, and deploy AI without falling into the ‘AI for everything’ pit or losing sight of the value that data governance and analytics bring.
It’s a canny strategy, particularly given how many of SAS’ customers are in heavily regulated industries and continue to approach AI with a degree of skepticism.
Preparing for IPO
In our preview piece for the conference, we speculated that SAS could provide an update on its long-held IPO ambitions, first announced in 2021 with a goal of going public in 2024. While the keynotes and press releases were absent of any further news on this, Goodnight gave ITPro some insight into SAS’ timeline and what being IPO-ready means for the firm.
“We had 14 financial systems around the world, we’ve now consolidated to one and it has taken us about three to four years to get that in place,” Goodnight said, explaining that this process was finalized in January 2025.
He added that SAS has also been getting lead to cash (L2C) systems, which encompass the entire process of acquiring customer leads to securing sales, in place as this is something public companies are expected to have.
“That went into operation on the first of the month,” he said. “So everything’s in place but now we have to learn to use this new system that no one’s familiar with, so they’re working overtime right now trying to feed quotes and stuff into the system. So we’ll need to run on that for at least a year before we’re ready.”
With this in mind, Goodnight said 2027 was the likeliest year for SAS’ IPO. Day has been heavily involved in the IPO preparation process, told ITPro that external economic factors were also at play here.
“A little bit over half of our revenue comes from outside of the United States, so we’ve had to pay attention to the global economy for a long time,” he said, adding SAS will continue to closely monitor the situation for the governments it works with.
“My job is to make sure we can operate as a public company, we can report as a public company, we have sustainable revenue growth, we have sustainable profit. When I do that, I give Dr Goodnight options, he can choose when to go.”
When it does finally happen, the SAS IPO will be an all-important milestone for the company, especially given the tight control Goodnight has maintained in his near 50-year reign as CEO.
Goodnight’s faith in the potential and longevity of SAS is evident in fact he’s turned down several acquisition offers in the past.
“I didn’t like the price,” Goodnight said of the offers he’s received in the past, adding that he was reluctant to see SAS absorbed into another firm.
In 2021, the Wall Street Journal reported that alleged Broadcom plans to acquire SAS for $15-20 billion had fallen through after Goodnight and co-founder John Sall changed their minds.
Quantum is here – with a catch
For some years, quantum computing has largely been discussed in terms of the potential risks it poses to common encryption algorithms like RCS. Cybersecurity experts have warned of ‘steal now, crack later’ attacks in which hackers exfiltrate evergreen encrypted data with the long-term goal of decrypting it when they eventually obtain a quantum computer.
At SAS Innovate 2025, the firm made something of a splash by claiming quantum is already here – and delivering results for its customers. Using what it calls quantum AI, SAS helped P&G to solve complex combinatorial problems, preventing ingredients from cross-contaminating products its manufacturing process.
As explained in the ITPro Podcast episode on the conference embedded below, this is not the same thing as the quantum computing often discussed in the industry. It was in fact the result of a process known as ‘quantum annealing’. This is essentially a very advanced, quantum-based optimization algorithm.
The term quantum AI seems to be doing some work for SAS here, blurring the line between bona-fide quantum calculations and the existing solver algorithms with which SAS is well-acquainted. Even Goodnight questioned the name, telling ITPro:
“I don’t know why they call it AI, it’s not really AI, but everything’s got to have AI on the end of it.”
Regardless of the semantics, SAS is investing in quantum solutions now to be ready for the significant future demand it’s predicting in the market. With SAS staying private for the time being and Goodnight telling ITPro that R&D has always been an important area of focus for the firm, it has room to expand its spending on quantum ahead of these projected returns.
Perhaps unsurprisingly, data underpins all the announcements and claims made at SAS Innovate 2025. As it nears a half-century in the analytics space, SAS appears more confident than ever that whatever the current technological trend, it has the systems and approaches to help any enterprise transform its data into tangible value.
Source link
