AI Strategy Lessons from IQVIA for Singapore Firms

US healthcare data and clinical research provider IQVIA just got a live-fire test of a question many Singapore leaders are quietly asking: will new AI tools replace what we sell, or make what we sell more valuable?

On 5 Feb 2026, IQVIA defended its AI strategy on an earnings call after analyst pushback and a market wobble that punished “software and services” names. The details matter: IQVIA forecast 2026 adjusted earnings of US$12.55–US$12.85 per share (below an average estimate of US$12.95), citing roughly US$80 million in higher interest expense, while projecting revenue of US$17.15–US$17.35 billion (above an estimate of US$17.07 billion). Shares fell more than 8% on the day.

This post is part of the AI Business Tools Singapore series—practical notes on how businesses here can adopt AI for operations, customer engagement, and growth without getting trapped by hype. IQVIA is a useful case study because it sits in a regulated, data-heavy industry where AI adoption is real, but careless AI use is expensive.

Source article (landing page): https://www.channelnewsasia.com/business/iqvia-backs-ai-strategy-analysts-question-impact-business-5910261

Why IQVIA’s AI debate matters to Singapore businesses

Answer first: IQVIA’s situation shows that AI disruption is rarely “AI replaces your company.” It’s usually “AI changes the value of what you already own”—especially proprietary data, workflows, and trust.

The analyst concern in the Reuters/CNA report wasn’t absurd. Over the last year, general-purpose AI tools (from vendors like Anthropic, among others) have moved from “chatbots” to credible substitutes for parts of consulting, analytics, and documentation-heavy professional services. That shift can spook investors because it compresses margins in businesses that look, from the outside, like knowledge work.

But here’s the stance I think is right for Singapore operators: most companies get the threat model wrong. They focus on whether an AI model can “do the work.” They ignore whether it can do the work with your constraints—your data rights, audit requirements, risk tolerance, and client obligations.

That’s why IQVIA’s CEO kept returning to a single point: proprietary healthcare information assets. You can debate tone (“really frustrating”), but the business logic is sound.

A Singapore-specific lens: regulation, procurement, and trust

Singapore is a place where AI adoption is accelerating, but it’s not a free-for-all. Many sectors—life sciences, healthcare, finance, critical infrastructure—operate under strict governance. In practice, that means:

• Buyers ask where data goes, who can access it, and how it’s retained.
• Security reviews and vendor risk assessments slow down “just try it” rollouts.
• Audit trails, traceability, and SOP alignment matter as much as model quality.

So the companies that win aren’t the ones with the flashiest model demos. They’re the ones that operationalise AI safely and can prove impact.

The real moat in an AI era: proprietary data + workflow integration

Answer first: AI models are becoming commoditised; data rights and workflow position are not.

When IQVIA argues its data assets “cannot be replicated by general-purpose AI models,” it’s making a broader point: if you own unique, legally usable datasets—and have the relationships and processes to apply them—AI becomes an amplifier.

For Singapore life sciences and healthcare-adjacent businesses, the analogue is straightforward:

• If your “asset” is a pile of PDFs, scattered spreadsheets, and staff knowledge, AI will help—but competitors can copy you.
• If your asset is clean, consented, governed data plus embedded processes (clinical ops, QA, regulatory, safety, supply chain), AI increases your defensibility.

What counts as “proprietary data” in practice?

It’s not just “we have data.” It’s data you can legally use, that others can’t easily assemble, and that is maintained over time.

Examples relevant to Singapore organisations:

• Longitudinal patient or trial datasets with clear consent and governance
• Pharmacovigilance case histories with consistent coding and metadata
• Operational datasets: deviations, CAPA histories, batch records, change controls
• Commercial datasets: HCP engagement history, formulary movement, tender outcomes

The unglamorous work—taxonomy, master data management, consistent identifiers—is what makes AI outputs reliable.

What “AI strategy” should look like (and where many companies slip)

Answer first: A credible AI strategy is a portfolio of 3 things—use cases, data readiness, and operating model—tied to measurable business outcomes.

IQVIA’s critics were essentially asking: “What if AI tools displace your services?” The better question for any Singapore leadership team is: which parts of our value chain are automatable, and which parts become more valuable when automated?

1) Use cases: start with high-frequency, low-ambiguity workflows

If you want AI ROI you can defend in a board meeting, start where volume is high and variance is manageable.

For life sciences and healthcare services firms, strong early use cases include:

• Document intelligence: summarising protocols, extracting endpoints, identifying inconsistencies
• Medico-legal review support: drafting first-pass responses, highlighting missing references (with human sign-off)
• Safety case triage: classifying incoming reports, routing, and suggesting coding (MedDRA/WHO-DD support)
• Commercial ops: account planning drafts, call note structuring, next-best-action suggestions

Avoid starting with “fully autonomous” anything. In regulated environments, that’s how pilots get paused.

2) Data readiness: get serious about permissions and provenance

Most AI programs stall because teams try to feed sensitive data into tools that weren’t procured for it.

A pragmatic checklist:

• Data classification: what is confidential, regulated, personal data, or export-controlled?
• Rights to use: consent, contracts, third-party terms
• Provenance: can you trace where a claim came from?
• Retention: is data stored, cached, or used for training?

If you can’t answer those, your “AI strategy” is a slide deck.

3) Operating model: decide who owns outcomes, not just models

IQVIA’s comments hint at an important truth: AI doesn’t sit in IT; it sits in the business.

A workable operating model in Singapore organisations typically includes:

• A business owner per use case (Ops, Medical, Quality, Sales)
• Central enablement (data, security, procurement, legal)
• Clear SOPs for human review and exception handling
• A measurement plan (time saved, error rate, cycle time, compliance outcomes)

How to evaluate “AI threat” the way analysts do (but more usefully)

Answer first: Treat AI as both a cost-down tool and a product/advantage tool—then quantify exposure and upside separately.

In the IQVIA story, the market reaction mixed two issues:

1. Business-model fear: AI could commoditise services.
2. Near-term financials: profit guidance was below expectations due to financing costs.

Singapore leaders should separate these too. Even if AI improves productivity, margins can still be squeezed if pricing collapses. The right response isn’t “avoid AI.” It’s reposition your offer.

A simple 2x2 that works in real life

Map each service line (or internal function) by:

• Substitutability: can a general AI tool do 70% of it with acceptable risk?
• Data advantage: do you have unique data/process position that improves results?

Actions:

• High substitutability + low data advantage: automate internally fast; expect pricing pressure; redesign roles.
• High substitutability + high data advantage: productise “AI + your data” as a premium offer.
• Low substitutability + low data advantage: build guardrails; improve data; don’t oversell AI.
• Low substitutability + high data advantage: invest; this is where durable differentiation lives.

That’s the IQVIA argument in operational form.

AI business tools Singapore companies can adopt now (without chaos)

Answer first: You don’t need a giant platform program to get value; you need the right AI business tools matched to controlled data and clear workflows.

Here are tool categories I’ve seen work well in Singapore rollouts, especially in regulated or enterprise settings:

AI copilots for internal knowledge (with access control)

Goal: reduce time spent searching SOPs, policies, and past decisions.

• Use role-based access control.
• Restrict sources to approved repositories.
• Log queries and outputs for review.

Document and contract intelligence

Goal: extract structured fields, track changes, and flag risks.

• Start with one document type (e.g., protocols, quality agreements, vendor MSAs).
• Define “gold fields” that must be accurate (dates, parties, obligations, endpoints).

Customer engagement support

Goal: improve consistency of responses and shorten turnaround.

• Draft emails and proposals from templates.
• Add a human approval step and tone guidelines.
• Measure response time and customer satisfaction.

Analytics acceleration

Goal: help analysts move faster without trusting AI blindly.

• Use AI for query drafting, chart suggestions, and narrative summaries.
• Keep the “source of truth” in your BI stack; AI should not become the database.

People also ask: “Will AI replace professional services in Singapore?”

Answer first: AI will replace tasks, not entire firms—unless the firm refuses to change packaging, pricing, and proof of value.

If your company bills by hours for repeatable deliverables, AI will pressure you. The survivable path is:

• Move to outcome-based pricing where possible
• Embed proprietary data, benchmarks, or governance into the service
• Build repeatable internal accelerators (templates, playbooks, automation)
• Prove compliance and quality, not just speed

IQVIA’s defense is a reminder: clients don’t buy “analysis.” They buy trusted decisions under constraints.

What to do next if you’re leading AI adoption in Singapore

Answer first: Pick one workflow, measure it hard, and build governance as you scale.

If you’re in life sciences, healthcare services, or any data-heavy business in Singapore, I’d start with this 30-day plan:

1. Choose a narrow workflow (e.g., trial document summarisation, safety intake triage, QA deviation drafting).
2. Define success metrics (cycle time, rework rate, audit findings, SLA compliance).
3. Lock down data rules (what can be used, where it runs, retention, access).
4. Run a pilot with humans in the loop and capture exceptions.
5. Decide whether to scale based on metrics, not enthusiasm.

The broader theme in the AI Business Tools Singapore series is simple: tools are easy to buy; capability is harder to build. IQVIA’s story is what capability-building looks like under scrutiny.

AI will keep improving, and skepticism will keep surfacing—especially after market selloffs and earnings misses. The practical question for Singapore businesses is: when AI gets cheaper and better, what part of your advantage becomes clearer rather than weaker?