50 Years and Still Going Strong: SAS Innovates in AI, Digital Twin, and Quantum Computing
A special edition of SAS Innovate 2026 has just concluded, the company’s event dedicated to partners and customers. Edge9 was present in Dallas to celebrate the fiftieth anniversary of the company founded in 1976 by James Goodnight, who still energetically leads an organization inspired by values not always shared by competitors in the industry.
The backdrop is typical of an enterprise conference by a major software vendor, but the substance differs from many other industry events. SAS remains a private, financially solid company, little concerned with trends. The way it presented its product announcements on AI is consistent with a mindset perceived in informal exchanges during panels and closed-door sessions: full awareness of its stature, no rush to chase the current messaging, an explicit claim of its own history. A calm strength, in other words.
The perspective is immediately framed by Jenn Chase, SAS CMO, in an interview: "Our values, shaped over the years by Jim Goodnight, focus on the problem or opportunity, not the technology. We work backward, starting from the problem, and apply any technology needed. It’s a bit of a contrarian view in today’s market." This same logic justifies the keynote's focus, where CTO Bryan Harris opened by asking, "Will people matter?" suggesting that the real crisis at the moment is not caused by the expansion of AI, but by the loss of confidence in human ingenuity.
Among those who shared their part of the announcements at Innovate 2026 is Marinela Profi, Global GenAI & Agentic AI Strategy Lead at SAS. This detail is noteworthy for the Italian audience: Profi was born in Rome and has worked at the headquarters in Raleigh for six years, in a global role that involves traveling and supporting SAS customers across all continents where the company is present. Her position holds responsibility over a key technology for the future of the platform in a leading enterprise software company: a trajectory that remains an exception for an Italian and merits recognition.
The Strategic Framework
Underpinning the announcements is a market trajectory that Harris, during the media briefing following the keynote, referred to as the commoditization curve. In the latest technological waves, each level of software has been eroded by its successor: from open source, which has made development a basic function, to hyperscalers that have trivialized storage, networking, and computing. Today, it’s the turn of generative AI, which is changing the economics of the build versus buy decision in enterprise applications. Customer companies are asking their vendors if that software still offers value proportional to its cost, and with agentic AI, an increasing number of functions can be replicated at low cost by generalist agents. The consequence, Harris argued, is that every software vendor must rethink its value proposition. SAS does this by investing in four pillars: AI governance, agentic AI, digital twin, and quantum AI.
Underlying these four pillars is a financial position that Harris indicated as enabling. SAS remains private, has been profitable for fifty years, and has no debts on its balance sheet. "You can think long-term only if you don’t have fifty or one hundred billion in debt to manage," he responded to a journalist who asked how the company reconciles short-term pressures with investments in multi-year horizons. It’s the same solidity that supports the human framework of the keynote, built around the question, "Will people matter?" Only the freedom from having to chase the market makes such an attitude sustainable.
Openness to External LLMs and Cross-Governance
The first two product announcements, the SAS Viya Model Context Protocol Server and SAS AI Navigator, are interpreted within this framework. Marinela Profi explained: "Last year we decided that the first thing to do regarding agentic AI was to invest in a dedicated architecture above Viya, so that every future feature originates from the same foundation, with governance, security, monitoring, and orchestration integrated from the start." This is why, she said, the company could move quickly while other market players that rushed forward without foundations are now managing their technical debts.
"From the outside, SAS may seem slower. After three years, we are, however, one of the few companies showing real agentic AI cases, quantum computing, and digital twin."
The MCP Server translates this choice into an explicit openness to commercial LLMs. SAS exposes the analytical functions, models, and decision automation logics of Viya as tools callable by agents built on external LLMs like Claude. This shift turns a historical line for the company on its head: the corporate analytical stack of 2026 is heterogeneous by definition, and the conversational agent no longer has to be proprietary as long as the resulting decisions pass through models and data under governance.
AI Navigator complements the MCP Server in terms of governance. Reggie Townsend, Vice President of SAS AI Ethics, Governance, and Social Impact, spoke about it in a press meeting. AI Navigator is a SaaS service launching in Q3 2026 on the Microsoft Azure Marketplace, which allows organizations to inventory and maintain governance over their existing AI use cases, underlying models, and applied policies. The strategic detail is the non-dependence on Viya. AI Navigator also governs models and agents that SAS did not develop, from Claude to Microsoft Copilot, applying the same rules that apply to the platform's internal models.
"The next chapter of AI must focus on bringing human judgment to large numbers, governing at the right speed, and making trust a competitive advantage. Judgment remains a human prerogative," Townsend stated. On the operational risk of governance, his statement is clear: "The biggest risk of an AI governance program is not regulation, but adopting such a complex tool that no one ends up using it."
Agentic AI for Regulated Processes
The vertical agentic AI initiative, presented within a billion-dollar plan for sector solutions, is the practical demonstration of the same logic. The main announcement is the SAS Supply Chain Agent, currently in private preview, which transforms the sales and operations planning process from a typical monthly cadence to continuous. The agent balances demand, availability, and logistics in real time and accepts hypothetical scenarios via chat. Alongside this, the first Industry Copilot solutions are already available, particularly an assistant for managing financial risks in banks and another for clinical data discovery in healthcare, with extensions on the roadmap for 2026 towards preventing financial crime in banking and optimizing planning in manufacturing. A data mapping agent, introduced last year, is now integrated into sector models for therapeutic adherence, automating the mapping of clinical data—a manual activity that often delays healthcare analytics projects.
In real cases, the logic remains the same: starting from the industry problem and applying the required technology. The State of Nevada and other U.S. states use a SAS solution to reduce errors in federally funded food assistance programs, avoiding penalties introduced by new regulations and improving subsidy delivery. Banks and insurers rely on a fraud decisioning agent for payments, trained on shared data through a consortium of large global financial institutions, covering credit cards, debit cards, ATMs, digital wallets, and newer schemes like money mules. On the industrial side, the SAS Worker Safety program combines digital twins built in Unreal Engine, synthetic data, and computer vision to train models that detect missing protective equipment, even in rare scenarios such as collisions between forklifts for which there is insufficient real video footage.
SAS summarized the logic in one sentence: the industrial context transforms generic AI into reliable AI, reducing complexity and time to achieve a return. The figure of scale is provided by the SAS Hackathon, now in its sixth edition, which this year involved 1,400 participants from over 120 countries, largely with cases that later went into production.
Digital Twin and Synthetic Vision
The program on digital twins is likely the one with the fastest development curve among the four pillars. SAS showcased at Innovate 2026 an evolution of the work done last year with Georgia Pacific, applied this year to an American supplier of sterilization for medical devices. The digital twin of the facility, built in Unreal Engine, revealed a bottleneck that management had misinterpreted. The logic is to be able to ask scenario questions about the operation of the plant at low cost, from doubling the machines to repositioning materials, which in the real world would require weeks and investments.
On this basis, SAS presented a technological innovation registered as patent-pending:Synthetic Vision, a method for training computer vision models entirely on synthetic data produced by digital twins. The advantages are threefold: modeling rare events for which real footage is sparse or absent, covering scenarios that are impossible to film, such as a person without protective equipment in a sterile area, and eliminating the need for manual labeling of frames—a time-consuming and costly activity. The application scope is not just industrial; SAS has demonstrated applications in tracking laboratory samples in healthcare, monitoring leaks and fires in the oil and gas sector, and inspecting power lines with drone footage in the utility sector. The technical stack combines Viya for analytical models and SAS Event Stream Processing for real-time inference at the edge.
Quantum Computing: Quantum AI as a Bridge
On quantum computing, SAS's approach illustrates the non-sensationalistic nature of the company best. Bill Wisotsky, Principal Quantum Systems Architect at SAS, explained the architecture of the new SAS Quantum Lab, coming in Q4 2026 for Viya customers. The problem the lab addresses is economic: running a single quantum algorithm on real hardware costs hundreds of thousands of dollars, and to get useful results, many variations must be tested. SAS reused the CAS workers of its analytical platform, meaning the distributed computing nodes that run Viya workloads, transforming them into parallelized quantum simulators. The customer builds and optimizes the algorithm in the simulated environment, achieving acceleration of 100-300 times compared to a conventional simulator, and only the best algorithm is actually sent to the real quantum hardware. The underlying logic is one of common sense and pragmatism, revealing much about the character of the company.
Choosing the quantum hardware supplier occurs downstream, based on what emerges from the simulation and the specific nature of the algorithm: it’s the SAS experts who indicate the most suitable vendor for the unique problem, drawing from a pool that includes D-Wave for combinatorial optimization, IBM, IonQ, and Quera for other classes of computation. Wisotsky shared an anecdote that serves as a manifesto of the positioning. An insurance company approached SAS for a quantum POC. SAS experts reframed the problem and solved it classically in one minute and thirty seconds, closing the POC without involving any quantum hardware. "If we want quantum to be useful, we need to stop playing with sensationalistically titled problems. When I look at academic publications, this kind of problem is decreasing; real cases are growing," he added. This approach resembles that of an old statistical analytics company rather than that of a cloud vendor, placing SAS in a different space from hyperscalers that sell direct access to quantum power.
Mid-market, Partnerships, and Business Model
Regarding the mid-market, a segment where SAS has historically been less present, Chase pointed to Viya Essentials as the new entry point. It’s an offering that brings the functionalities of Viya into a cloud directly managed by SAS and is designed for businesses that cannot afford to manage the platform independently. The company also confirmed an investment in local partner channels in European markets like Italy, where direct presence is less widespread. On cloud partnerships, alongside the three hyperscalers comes OVH Cloud, selected to address European requirements for digital sovereignty. Hyperscalers and large data platforms remain an enabling ecosystem, while the real direct competitors, Chase noted, are companies like Databricks and Palantir.
On the business model, when specifically asked about the arrival of Viya as a multi-tenant service on the marketplaces of the three hyperscalers, the company responded cautiously. The internal conversation has expanded beyond consumption-based pricing to include hypotheses of results-based pricing. Workbench on the Service, announced at Innovate 2026, is described as a first concrete step towards a more complete SaaS offering. The multi-tenant architecture of Viya includes three distinct layers: control, application, and data, which can be managed by SAS, the customer, or in a hybrid mode. The data plane remains a sensitive point for customers in regulated industries: in the current version of Workbench, it was the customer's responsibility; in the new version, it is SAS's responsibility.
What remains from SAS Innovate 2026, in the end, is less a list of products than a confirmation of identity. The question "Will people matter?" from the CTO is not marketing rhetoric; it’s how SAS states its unwillingness to sell AI that sidesteps human judgment. The reuse of CAS workers as quantum simulators, the rewriting of governance architecture before features, the choice to solve a problem classically instead of forcing it into quantum are episodes that confirm the same logic. For the enterprise audience, the message is clear: SAS is not chasing AI; it is structuring it within a business method of analysis that has existed for fifty years. The amount of work needed to get there explains why the company may seem slower from the outside. It also explains why, after three years of widespread generative AI, it remains one of the few able to show real cases in production and not just slides of prototypes.