US healthcare's structural rewrite — and why it took twenty years.

Sudhir's opening frame: there is no other industry where a payment model invented in 1965 still runs the cash flow, an interoperability standard set in 1996 still gates the data, and a coding system from the 1970s still determines whether a claim is paid. Healthcare moves slowly not because it is technically backward but because the regulation underneath every workflow is a thirty-year sediment that nobody can rip out at once.

That slowness is itself a competitive feature — most consumer-software companies cannot tolerate the cadence. The firms that can are the ones that started building inside the regulation rather than around it. CitiusTech is one of those.

Sudhir keeps returning to it. Fee-for-service pays a hospital for what it does: every imaging study, every consult, every admission. Value-based care pays it for what happens to a population over a period: the cost of care, the readmission rate, the HEDIS score. Change the unit of payment and you change every system that reports against it. The EHR has to capture outcomes, not just charges. The claims platform has to attribute members to providers. The analytics has to think in cohorts and member-years, not encounters.

The transition is twenty years old in policy and barely ten years old in production systems. Most of the rewrite is still in front of the industry. That is the order book.

CitiusTech's revenue map is a three-legged stool. Payers — the insurance companies, the Blues plans, the Medicare Advantage runners — need claims engines, member analytics, prior-authorisation automation, and the increasingly heavy machinery of HEDIS and STAR-ratings reporting. Providers — the IDNs, the academic medical centres, the ACOs — need EHR integration, revenue cycle management, clinical analytics, and increasingly the ambient-scribing rails. Life sciences — pharma, medical devices — need real-world evidence pipelines, clinical-trial data infrastructure, and compliance tooling that satisfies the FDA and EMA.

The three pillars do not buy the same software. They buy adjacent skill sets from the same firm. The diversification is the moat: a downturn in payer IT spend is rarely synchronous with life-sciences procurement.

Roughly two-thirds of US hospital beds run on Epic. Sudhir is precise about why: it is not that Epic's database design or UI is better. It is that an Epic install touches every workflow in the hospital — registration, scheduling, orders, results, billing — and the cost of switching is measured not in licence fees but in retraining tens of thousands of clinicians who learned medicine through that interface. The moat is the muscle memory of a generation of doctors.

That is also why the third-party-app layer above Epic is where the interesting product opportunity lives. Replacing Epic is a fool's errand. Building tools that sit on top of Epic — through its APIs, through FHIR — is a quarter-long integration rather than a year-long migration. CitiusTech lives in that adjacency.

Oracle's twenty-eight-billion-dollar acquisition of Cerner is one of the largest healthcare-IT deals in history and the press has mostly stopped covering it. Sudhir's read: Oracle is in the middle of replatforming Cerner onto its cloud, rebuilding the data model, and trying to win back the share that has drifted to Epic. Whether that bet pays off shapes the second-most-important EHR in the country for the next decade.

The under-discussed implication is the data architecture. Oracle is the only EHR vendor with a serious bet on cloud-native, FHIR-first design at the foundation rather than as a translation layer. If they pull it off, the analytics ceiling on the Cerner footprint rises sharply. If they don't, Epic keeps consolidating.

From outside, the alphabet soup looks like overhead. From inside, Sudhir argues, it is the qualifier and ultimately the protection. HIPAA gave the industry its data-handling rules in 1996. HITECH in 2009 forced electronic adoption with carrots and sticks. SOC2 became the audit grammar customers learned to ask for. The result is that any firm doing serious work in US healthcare has been writing the same kinds of attestations and runbooks for fifteen years.

The interesting structural fact is that this turns into a moat in the opposite direction from how it looks. The first audit is expensive. The hundredth is a process. Firms that paid the early cost now compound on it; new entrants discover that compliance-from-scratch costs more than the differentiated engineering work they came in to do.

Fast Healthcare Interoperability Resources — FHIR — is the API standard that lets one healthcare system talk to another. It was meant to fix the absurdity of an industry that could not move a patient's record across the street. For most of its existence FHIR was a slide in a keynote, not a wire-level reality. Sudhir's view is that the last three years have changed that: CMS mandates, the 21st Century Cures Act, and finally an EHR vendor base that ships usable FHIR endpoints have crossed a tipping point.

The downstream implication is large. Once data flows out of the EHR through a standard API, the analytics, the AI, the third-party tools all become unblocked. Twenty years of "data trapped in the EHR" complaints become a transition rather than a permanent state.

Sudhir is unsentimental about the AI hype cycle in healthcare. Most of what was announced in 2023 was a chatbot bolted onto a clinical workflow nobody redesigned. Ambient scribing — Abridge, Nuance/Microsoft DAX, a handful of others — was different. The model listens to the encounter, drafts the note, files it in the EHR. The clinician sees a finished document instead of typing through dinner.

It worked because it solved the right unit of value. Doctors do not need help reasoning; they need their evenings back. The win is two hours of pajama-time documentation removed from every weeknight. That number is so visible to a clinician's life that the buying decision moves out of IT and into the chief medical officer's office. CitiusTech's bet is that this pattern — AI that returns hours, not AI that replaces tasks — is the template for the next ten years of clinical software.

Once payment shifts to outcomes over a panel, the question changes. Instead of "did this patient get the right care during this admission," the question is "which of my forty thousand attributed members are rising-risk this quarter, and what is the cheapest intervention that keeps them out of the hospital." That is a different software product. It needs claims, EHR, social determinants, pharmacy fills, and increasingly device telemetry, all stitched and risk-stratified.

Sudhir notes that the analytics surface is unevenly developed. Payers are further along — they have always been data companies in disguise. Providers are catching up under ACO pressure. The interesting middle is the joint venture: payer-provider entities like Optum, Kaiser, the integrated delivery systems, where the data sits on one side of the firewall and the financial accountability on the other.

One in three US physicians reports burnout at any given moment, and a meaningful share are leaving practice or moving to part-time work. The proximate cause is documentation: hours of EHR clicks per hour of patient time. Sudhir frames this as the most under-appreciated design constraint in the industry. Every new product, every new analytics dashboard, every regulatory check-box adds to a workflow that is already at breaking point.

The reframe is that the best healthcare-IT products of the next decade will be measured by what they remove from the clinician's day, not what they add. Ambient scribing succeeds on this axis. So does prior-authorisation automation. So does retrospective coding. The market for "add another button to Epic" is functionally closed.

The Centers for Medicare and Medicaid Services — CMS — sets the cadence of half the industry. Pricing-transparency rules in 2021, the interoperability rule, the No Surprises Act, the prior-authorisation rule, Medicare Advantage payment-model updates each year. Every CMS rule is, downstream, a software project at every payer and every provider. Sudhir's working observation: CitiusTech's project pipeline lags CMS rule-making by about nine months.

This is a feature of US healthcare, not a bug. The biggest customer of US healthcare is the US government. The government's policies are the product roadmap for the industry that serves the government. A firm that reads CMS rules carefully has a leading indicator on its own demand curve that most consumer-software firms cannot dream of.

Sudhir reaches for the analogy with measured confidence. India is the back office of US healthcare in the same way it became the back office of US banking in the 2000s — Cognizant TriZetto, Wipro, TCS, Sutherland, and CitiusTech itself are no longer doing low-end ticket work. They are running the clinical-trial analytics for top-five pharma. They are operating the claims systems for major Blues plans. They are writing the ambient-scribing pipelines.

The under-told fact is that this took twenty years to compound and is now its own category. There is a stack of skill — domain knowledge, compliance experience, HL7 and FHIR fluency, ICD-10 coding literacy — that simply does not exist outside India and a handful of US specialists. The GCC (global capability centre) model, which Indian engineers built in financial services, is replicating in healthcare in the late 2020s.

CMS now requires hospitals to publish negotiated rates and insurers to publish their pricing files. The data is messy, inconsistent, and not yet wired into the consumer experience the rule-makers imagined. Sudhir is patient about the implication: the second-order effects, not the first. The data exists. Aggregators are cleaning it. Employers are starting to use it in benefits design. Reference-based pricing is creeping back. The cost-opaque era of US healthcare is ending — not next year, but inside this decade.

The downstream IT consequence is large. Every payer has to support transparency-driven workflows. Every provider has to defend its rates in negotiation. The analytics tooling that lets you compare reimbursement across markets is a new category that did not exist five years ago.

Half of US Medicare beneficiaries are now on Medicare Advantage plans rather than traditional Medicare. That single fact is reshaping the payer landscape. MA plans are paid a capitated rate per enrolled senior; their margin lives in care management, STAR ratings, and risk-adjustment accuracy. Sudhir treats this as the single biggest driver of payer IT spend over the next five years.

The downstream demand is for software that is good at exactly the things MA plans live and die by: HEDIS measure capture, STAR uplift programmes, risk-adjustment factor (RAF) coding, member engagement analytics, and care-management workflow. That stack is being rebuilt right now. Every regional Blues plan is buying it or building it.

Sudhir's closing frame is restrained. He does not promise revolution. He promises that the changes already underway will finish — that the value-based-care share of payment will keep climbing, that ambient scribing will become table stakes, that FHIR will be the default API, that population-health analytics will be a budgeted line item at every provider above a certain size. None of these are predictions about the next product. They are predictions about completion.

The thing that ages well in this conversation is the discipline of incremental claims. Healthcare rewards firms that stay in the room for two decades and bet on completion. The exciting bets are the ones the keynote stage announced ten years ago and are quietly turning real now.