In a business that adores a new label, Persistent Systems has survived by keeping an old job. It builds software that cannot be allowed to wobble. Bank ledgers, clinical data systems, cloud platforms, telecom machinery and the hidden layers beneath commercial software are its natural habitat. The work is specific, frequently anonymous and difficult to photograph. It is also the work that decides whether a clever technology becomes an operating business or an expensive demonstration.
That makes Persistent unusually legible in the AI moment. Companies already have models. What they often lack is clean data, usable APIs, governed workflows and a sensible way to connect a probabilistic machine to systems that move money, manage patients or run production. Persistent sells the engineering between the promise and the working system.
The Pune company is no overnight AI conversion. Anand Deshpande, a computer scientist who had worked at Hewlett-Packard Laboratories, incorporated Persistent in 1990 and returned to India later that year. The first office occupied about 350 square feet in Pune's fledgling Software Technology Park. The name itself came with a programmer's wink: persistent data is the kind that remains after a program ends or a machine shuts down. A durable object was the aspiration before durability became a corporate metaphor.
The company beneath the company
Persistent began close to the database. Early company histories place Hewlett-Packard among its customers by 1992-93 and Microsoft by 1995-96. This was product engineering: helping technology companies design, test, extend and maintain the software they sold. The distinction matters. Traditional outsourcing can be organized around taking cost out of a known process. Product engineering begins with a messier question - what should the product become, and how can it keep working while it changes?
Today the catalogue is much broader. Persistent advises on technology strategy, develops and modernizes applications, moves infrastructure to public and hybrid clouds, connects enterprise systems, organizes data, redesigns customer experiences, secures identity and runs managed operations. It applies those skills across banks, insurers, healthcare providers, life-sciences companies, manufacturers, software vendors, telecom operators and consumer businesses.
Customers can hire it to rebuild a single platform or take responsibility for a wider estate. Recent disclosed wins include modernizing more than 250 applications for an insurance-services company, consolidating 3,000 software bots across 350 healthcare processes, and reengineering a multinational bank's trade ledger with generative AI. The names are often withheld. In enterprise technology, anonymity is not automatically evasive; it is frequently the price of touching systems that clients regard as strategic or regulated.
“The differentiator will not be the model itself, but the ability to create a unified Enterprise Context.”Sandeep Kalra, CEO, on the Q1 FY2027 results
That “context” is Persistent's neatest explanation of its current pitch. Its framework has three parts: core systems, business context and coordinated intelligence. Strengthen the applications and data underneath; capture the rules and history that give the data meaning; then let AI act across the organization with controls. It is not as instantly gratifying as a chatbot. It is far more useful when an answer needs to survive an auditor.
Services that learn to become products
Persistent is primarily paid for people and outcomes: multi-year engineering programs, consulting, cloud migrations, modernization projects and managed services, sold through fixed-price or time-and-materials contracts. Yet it is trying to make that labor more repeatable. Its own platforms act as packaged memory from previous engagements.
SASVA applies AI across the software-development life cycle. iAURA is aimed at data and analytics work. GenAI Hub lets enterprises develop and monitor applications across multiple large language models, build agent workflows, evaluate responses and watch costs. Around them sit narrower solutions for cybersecurity investigation, document intelligence, regulatory work, manufacturing and life sciences. These tools are not a clean break into subscription software. They are accelerators: software that helps Persistent deliver services faster, with more consistency and potentially better margins.
The model creates a useful loop. Engineers encounter the same expensive problem across clients. The company packages part of the solution. The next engagement begins with more than a blank screen. New client work produces another round of learning. Accenture and the Indian IT majors run versions of this play, while engineering specialists such as EPAM and Globant sell comparable depth. Persistent's argument is that its product-engineering roots give it a sharper feel for software architecture, while its smaller scale preserves more of a specialist's attention.
Engineering capacity, domain knowledge and accountability for turning old systems into modern ones.
AWS, Microsoft, Google, Salesforce, Databricks, Snowflake and others provide the platforms Persistent adapts.
Reusable methods, accelerators and experience with the awkward joins between systems.
Services remain people-heavy, partner-dependent and vulnerable to slow client spending or difficult integrations.
A partner ecosystem, not a private kingdom
Persistent does not ask an enterprise to throw away its technology choices. It works inside them. The partner list includes AWS, Microsoft, Google Cloud, IBM, Salesforce, ServiceNow, Databricks, Snowflake, NVIDIA, Red Hat, UiPath and others. A 19-plus-year Salesforce relationship, for example, has produced more than 1,500 joint engagements. In 2026, Persistent and Kong announced a partnership around the control layer for APIs, data, models and AI agents.
This vendor fluency is a strength and a constraint. It makes Persistent useful to a chief technology officer staring at six clouds and twenty years of acquired software. It also means the company competes inside ecosystems whose owners can change incentives, build adjacent services or favor another integrator. The response is breadth without total neutrality: become highly certified on the large platforms, then differentiate in the engineering that connects them.
The customer roster shows where that position lands. Persistent says it serves 20 Fortune 50 companies and four of the five largest banks in both the United States and India. Healthcare and life sciences are another deep pool, where it works on clinical data, laboratory systems, patient access and regulated analytics. Software and hi-tech clients still connect the modern company to its original trade: build the product, keep it alive, and modernize it without frightening the customers already using it.
Growth, then a much bigger bite
The numbers have rewarded the positioning. Fiscal 2026 revenue reached $1.654 billion, up 17.4 percent, with a 15.6 percent operating margin before interest and tax. In the quarter ending June 2026, revenue was $452.4 million, the twenty-fifth consecutive quarter of growth. Total contract value reached a quarterly record of $1.146 billion, helped by a 6.5-year services agreement worth more than $650 million with an unnamed global technology company.
Then came the conspicuous move. In June 2026, Persistent signed a business combination agreement with Nagarro and announced its intention to make an all-cash offer of €81 per share. Persistent had secured an agreement for roughly 21 percent of the German-listed engineering company; completion still depended on an acceptance threshold, approvals and the formal takeover process. If completed on the announced terms, the combination would create a group with roughly $2.9 billion in revenue, more than 46,000 employees and operations in over 40 countries.
The logic is geographic as much as technical. Persistent is strong in North America and wants greater European reach. Nagarro brings European business, enterprise-resource-planning and customer-experience capability, complementary industries and about 18,500 employees. The risk is equally plain. Buying scale is easier than integrating delivery cultures, client relationships and margins. Persistent's “boutique mindset” is appealing precisely because boutiques do not normally contain 46,000 people.
Where it fits
Persistent sits between three familiar categories. It is too implementation-heavy to be a pure strategy consultancy, too engineering-led to be understood only as a general outsourcer, and too service-oriented to be valued as a conventional software company. That middle can be awkward in a sales deck. In the market, it is useful. Enterprise technology problems do not respect category lines.
A client can use Persistent to decide which applications should survive, move them to cloud infrastructure, reorganize the data, add security, redesign the customer interface, introduce automation and continue operating the result. The company makes its best case when the assignment crosses boundaries and failure is expensive. It makes a weaker case when a buyer wants only low-cost staffing, a single off-the-shelf product or a famous strategic adviser in the boardroom.
Culture is part of the promise and part of the test. Persistent names four values - ingenious, confident, responsible and persistent - and emphasizes learning, inclusion, well-being and employee ownership. Those ideas must operate across 21 countries, a hybrid workforce and, potentially, a major acquisition. For a company that sells continuity while systems change, its own organizational modernization is more than an internal matter. Clients will be watching the demonstration.
The AI era has made Persistent's old specialty easier to explain. Models are abundant. Enterprise context is scarce. Turning one into the other requires patient, domain-specific engineering, followed by maintenance after the launch party leaves. Persistent has been practicing that routine since databases were the exciting new thing. The technology changed. The job persisted.