Why a pill and a pacemaker take different paths to your body

Why "the molecule is the product" changes the game

For a drug, you cannot see the risk. You cannot inspect a molecule's safety by looking at it. You have to give it to people and watch what happens. That is why drugs lean so heavily on large human trials.

For a device, much of the risk is visible and testable in a lab. You can stress test tubing, drop the housing, and audit the code. Some risk still requires human data, but a lot can be proven on a bench.

Different paths through the FDA

In the United States, the Food and Drug Administration (FDA) is the regulator, but drugs and devices go through separate doors.

The drug path: prove it works, molecule by molecule

A new drug typically moves through phases:

• Phase 1: small group, mainly safety.
• Phase 2: does it seem to work, and at what dose.
• Phase 3: large trial to confirm benefit and catch rarer harms.

Then the company files a New Drug Application (NDA) or, for biologics like insulin, a Biologics License Application (BLA). This process is slow and expensive because the core question ("what does this molecule do inside a human body?") can only be answered by dosing many humans.

The FDA's own plain-language overview is a solid starting point: How Drugs are Developed and Approved.

The device path: match a benchmark or prove novelty

Devices are sorted by risk into three classes:

• Class I (low risk, like a tongue depressor): light oversight.
• Class II (moderate risk, like many insulin pumps): often cleared through the 510(k) pathway, where you show your device is "substantially equivalent" to one already on the market.
• Class III (highest risk, like an implantable pacemaker): usually requires Premarket Approval (PMA), the most rigorous device review, with clinical data.

The 510(kkThe average number of new users each existing user generates through referrals. Above 1.0, growth compounds on itself and becomes exponential.View full definition →) route is why devices can iterate faster than drugs. If your new pump is meaningfully similar to an approved one, you may not need a giant new trial. A brand new drug almost never gets that shortcut.

Different risk, different lifespans

The risk profiles are not just different in kind. They are different across time.

A drug's risk is often front-loaded and chemical. Once a stable formulation is approved, each manufactured batch is chemically the same. The molecule does not "update."

A device's risk stretches across its working life. A pump can clog, a battery can die, a sensor can drift, and increasingly, software can have bugs. Modern devices ship software updates the way your phone does, which means the FDA now reviews how companies manage cybersecurity and post-market changes. A connected insulin pump that talks to a phone app is also a potential target for hacking, a risk a vial of insulin simply does not have.

This is why device companies live and breathe post-market surveillance (tracking problems after launch) and issue recalls. Drugs get recalled too, but device recalls often involve a physical fix, a firmware patch, or a replacement unit.

Different business models

Because the science and rules differ, the money works differently.

Drugs: patents, cliffs, and biosimilars

A drug's value is largely locked in its patent and its regulatory exclusivity. When those expire, competitors can sell copies. For small-molecule pills, those copies are generics, and they can crush the original's price fast.

For biologics like insulin, the copies are called biosimilars, not generics. Because biologics are grown in living cells, you cannot make an identical molecule, only a highly similar one. Proving that similarity is harder and costlier than making a generic pill, so biosimilar competition tends to arrive more slowly and cut prices less sharply. That nuance shaped years of debate over insulin affordability.

Devices: iteration, service, and switching costs

Device value is less about a single patent cliff and more about continuous improvement and lock-in. Once a hospital buys a fleet of pumps, trains staff, and integrates the software, switching to a rival is painful and expensive. Revenue often comes from consumables and service, not just the hardware. An insulin pump maker may sell the pump modestly but earn steadily from the disposable infusion sets, reservoirs, and sensors patients need every few days. That "razor and blades" model is common in medtech and rare in pharma.

When the line blurs: combination products

Real products increasingly refuse to stay in one box.

An insulin pen is a device (the pen) delivering a drug (the insulin). An automated insulin delivery system, sometimes called an "artificial pancreas," combines a glucose sensor, an algorithm, and a pump to dose insulin automatically. Is that a drug or a device? It is a combination product, and the FDA has a dedicated office to decide which rules dominate based on the product's primary function.

This matters commercially. A company that historically made molecules may suddenly need software engineers, cybersecurity teams, and hardware supply chains. A device company may need to understand pharmacology. The talent, the culture, and the risk tolerance differ, and mergers between these worlds often struggle for exactly that reason.

Software as a medical device

There is now a fully software-only category: Software as a Medical Device (SaMD). An app that analyzes glucose trends and recommends action can itself be regulated as a device, with no hardware and no molecule at all. As artificial intelligence enters diagnosis and dosing, regulators are working out how to oversee tools that keep learning and changing after approval. This is one of the most active frontiers in the sector for 2026.

How to reason about any biotech or medtech product

When you meet a new company or product, ask three questions:

1. Is the core value a molecule, a machine, or software? That predicts the science and the regulatory door.

2. Where does the risk live: inside the body's chemistry, in the hardware, or in the code? That predicts the testing and the recalls.

3. Where does the money come from: exclusivity, or lock-in and consumables? That predicts the business model and what competition will do.

Run insulin and the insulin pump through those three questions and the entire contrast falls out cleanly.

Key Takeaways

• Drugs are chemistry and biology; devices are physics and engineering. That root difference drives their separate regulatory paths, risks, and economics.
• The FDA uses different doors: drugs go through NDA or BLA after phased human trials, while devices are risk-classified and often cleared via 510(kkThe average number of new users each existing user generates through referrals. Above 1.0, growth compounds on itself and becomes exponential.View full definition →) or approved via PMA.
• Risk lives in different places over time: a molecule's danger is largely chemical and front-loaded, while a device faces mechanical, software, and even cybersecurity risk across its whole life.
• Business models diverge: drugs depend on patents and face generic or biosimilar cliffs, while devices rely on iteration, switching costs, and recurring consumable revenue.
• The boundary is dissolving: combination products and Software as a Medical Device force companies to master multiple disciplines at once, which is where much of the sector's future value and difficulty sits.