How a Peptide Drug Gets Made — and How AI Is Rewriting the First Step
How is a peptide drug like semaglutide made? A guide to solid-phase peptide synthesis, purification, and how AI tools like RFpeptides are rewriting step one.
Building a peptide is a chain reaction, one amino acid at a time. The chemistry is sixty years old; the way we decide what to build is changing fast.
Behind every GLP-1 injection is a manufacturing feat most patients never think about. Understanding how peptide drugs are made demystifies both why these medicines went into shortage and why the field is moving so quickly. The story has two halves: an old, elegant way of building peptides, and a new way of designing them.
Building the chain: solid-phase synthesis
The foundation was laid in the 1960s by Bruce Merrifield, who won the 1984 Nobel Prize in Chemistry for it. His method, solid-phase peptide synthesis, remains the backbone of the industry.
The idea is beautifully practical. Anchor the first amino acid to a tiny solid bead. Add the next amino acid; it bonds to the first. Wash away the leftovers. Add the next; wash again. Repeat, one residue at a time, until the full chain is built — then cleave the finished peptide off the bead. Because the growing chain stays fixed to the bead, you can flush out impurities at every step without losing your product. Each cycle is a small, controlled reaction, stacked into a precise sequence.
For a drug like semaglutide, the process is even more elaborate because the molecule includes an engineered fatty acid chain. This modification lets it last about a week in the body. That addition, and the purification it demands, is part of why peptide manufacturing is genuinely hard.
Purifying and finishing
A raw synthesized peptide is not a drug yet. It must be purified to pharmaceutical standards, stripping out the incomplete chains and byproducts that every synthesis produces. Then comes fill-finish: loading precise doses into pens or vials under sterile conditions, at industrial scale.
This is the unglamorous bottleneck. Fill-finish capacity, not chemistry alone, is part of what limited supply during the GLP-1 shortages, and it is a major reason Lilly and Novo are pouring tens of billions into new manufacturing plants. A drug you cannot fill into a pen is a drug patients cannot use.
The small-molecule shortcut
Not every GLP-1 drug is a peptide, and the difference shows up in the factory. Orforglipron is a small molecule, produced by conventional chemical manufacturing that is cheaper and simpler than peptide synthesis. That production advantage is one reason it launched at low prices — and a hint that the industry may lean on small molecules to serve the largest patient populations.
Rewriting the first step: AI design
Here is where the story turns futuristic. The hardest part of a peptide drug was never building it — it was deciding what to build — historically, finding a peptide that binds a target meant screening enormous libraries and hoping.
Tools like RFpeptides, from David Baker’s Nobel-winning lab, are collapsing that step. Described in Nature Chemical Biology in 2025, the system designs peptide binders from a target’s structure using deep learning. Chemists synthesize the winners with the same solid-phase method Merrifield pioneered. Design that once took years can now take days. The building technology is decades old; the design technology is brand new, and the two together are what make this an era rather than a moment.
The bottom line
A peptide drug is built like a molecular chain reaction — amino acid by amino acid, on a bead, then purified and filled with painstaking care. That manufacturing is difficult and capital-hungry, which explains both past shortages and today’s factory boom. What is changing fastest is not how peptides are assembled but how they are conceived: AI now drafts the molecule, and sixty-year-old chemistry brings it into the world—old craft, new blueprint — that is modern peptide medicine in one line.
Frequently asked questions
How is semaglutide manufactured? It is built through solid-phase peptide synthesis — assembling the amino-acid chain one residue at a time on a solid bead — plus the addition of an engineered fatty-acid chain that extends its half-life, followed by purification and sterile fill-finish into pens or vials.
What is solid-phase peptide synthesis? A method invented by Bruce Merrifield in the 1960s (Nobel Prize, 1984). The growing peptide stays anchored to a bead while amino acids are added and impurities washed away at each step, then the finished chain is cleaved off. It remains the backbone of the industry.
Is AI actually used to make peptide drugs? AI tools like RFpeptides now design candidate molecules from a target’s structure, collapsing discovery from years to days. Chemists then synthesize the winners using the same decades-old solid-phase method: old craft, new blueprint.
Sources
- Merrifield, R.B. “Solid Phase Peptide Synthesis. I. The Synthesis of a Tetrapeptide.” Journal of the American Chemical Society, 1963 (Nobel Prize in Chemistry, 1984).
- Rettie, S.A., et al. “Accurate de novo design of high-affinity protein-binding macrocycles using deep learning.” Nature Chemical Biology, 2025. DOI: 10.1038/s41589-025-01929-w.
- Muttenthaler, M., et al. “Trends in peptide drug discovery.” Nature Reviews Drug Discovery, 2021.
- Eli Lilly and Novo Nordisk — U.S. peptide manufacturing investment announcements, 2025–2026.