The Synthesis Gap: Why the Distance Between Peptide Design and Peptide Manufacturing Is the Biggest Bottleneck in Drug Discovery

Introduction

There is a particular kind of frustration that runs through modern peptide drug discovery programs, one that does not show up cleanly in project timelines or failure reports but shapes the pace of nearly every program it touches. A computational chemist or structural biologist identifies a peptide sequence with strong binding predictions and a compelling therapeutic rationale. The design is sound. The target is validated. The next step, in principle, is synthesis.

In practice, that next step frequently stops there.

The peptide is too long for conventional solid-phase synthesis. Or it contains a structural feature that existing vendors cannot reliably produce. Or the sequence has a run of hydrophobic residues that cause aggregation on the resin, killing yield and purity. Or it needs to be cyclic, or branched, or built from D-amino acids, and nobody in the supply chain can make that happen on a timeline that keeps the program moving.

This is the synthesis gap: the growing distance between what peptide designers can conceive and what peptide manufacturers can reliably deliver. It is not a new problem. But it has become a more consequential one, for reasons that are fundamental to where the field is heading.

Understanding the synthesis gap, where it comes from, how it is widening, and what it costs drug discovery programs in real terms, is increasingly important for anyone making decisions about peptide therapeutic development. It is also the problem that Amide Technologies, the Cambridge-based biotech developed out of MIT’s Pentelute Lab, was built specifically to solve.

What the Synthesis Gap Is, and Where It Comes From

The synthesis gap is not a single failure mode. It is a structural mismatch between two different timelines of scientific progress.

On the design side, the tools available to peptide drug designers have improved dramatically and continue to do so. Computational modelling, cryo-EM structural data, machine learning-assisted sequence optimization, and AI-driven de novo design have all expanded the range of peptide candidates that researchers can rationally propose. The designer’s toolkit is faster, more powerful, and more capable of generating structurally complex candidates than it has ever been.

On the manufacturing side, the dominant production methods have not kept pace. Solid-phase peptide synthesis (SPPS), which remains the standard approach for synthetic peptide production, was developed in the 1960s. It works reliably for short, relatively simple sequences. It does not work reliably for long sequences, structurally complex architectures, or many of the specific amino acid combinations that modern drug design increasingly favors. Biological expression covers different ground but has its own hard constraints, particularly for sequences that include non-natural amino acids or structural features that the ribosome cannot produce.

The result is a gap in the middle of the peptide drug discovery pipeline, a category of candidates that are designed, validated at the computational level, and then stranded because no reliable path to physical material exists.

This stranding is not always obvious. Programs do not typically announce that they have abandoned a candidate because of synthesis inaccessibility. The failure is recorded as a delay, a reformulation decision, a pivot to a different chemical series, or simply a candidate that never advanced from early-stage modeling to in vitro testing. The synthesis gap is often invisible in the language of drug discovery, even when it is the actual cause of program stall.

Amide describes this middle territory precisely: peptides too long for traditional SPPS, which caps out at roughly 40 amino acids with reliable quality, and impossible to obtain through biological expression because they contain mirror-image amino acids, branched structures, or other non-natural features. This is not a niche category. It is where a significant portion of the next generation of peptide therapeutics lives.

The Three Dimensions of the Synthesis Gap

The synthesis gap manifests across three distinct dimensions, each of which creates a different kind of blockage in a drug discovery program. Understanding them separately clarifies why there is no single technical fix and why addressing the gap requires a platform-level rethinking of how peptides are made.

Dimension 1: Length

Length is the most straightforward dimension of the synthesis gap and the one with the clearest numerical boundary. Conventional SPPS works by building a peptide chain one amino acid at a time on a solid resin support. Each coupling step is efficient but not perfect. Over a short sequence of 20 to 30 amino acids, the cumulative effect of sub-perfect efficiency at each step is manageable. Over a sequence of 60, 80, or 100 amino acids, the same coupling efficiency produces a mixture increasingly dominated by deletion sequences, truncations, and aggregated material. Yield and purity collapse.

The practical ceiling for reliable, high-quality SPPS under conventional conditions is approximately 40 amino acids. This number is not a regulatory threshold or an industry convention; it is a chemical reality that falls out of the mathematics of stepwise synthesis.

Many of the peptide candidates that are most interesting from a therapeutic standpoint are substantially longer than 40 amino acids. Mini-proteins, stapled peptides with extended scaffolds, long-chain cyclic peptides, and sequences designed to engage large protein-protein interaction surfaces frequently require 50 to 120 amino acids to achieve their structural and functional requirements. Under conventional synthesis, these candidates are effectively inaccessible.

Amide’s AFPS platform addresses this dimension directly, extending reliable linear synthesis to 120 amino acids through the use of heat and flow conditions that maintain coupling efficiency across longer chains. The four-day turnaround time for complex peptides, versus the weeks or months typical of conventional synthesis at comparable length and complexity, reflects what becomes possible when the length ceiling is meaningfully raised.

Dimension 2: Structural Complexity

Length alone does not define the synthesis gap. The second dimension is structural complexity: the range of architectural features that a peptide candidate may require and that conventional synthesis cannot reliably produce.

Several structural classes fall into this category. Highly hydrophobic sequences, where extended stretches of non-polar residues cause the growing peptide chain to aggregate on the resin before synthesis is complete, represent a persistent failure mode in standard SPPS. Cyclic peptides, which require head-to-tail or side-chain cyclization after linear assembly, add a post-synthesis step that is technically demanding and sequence-dependent in its success rate. Branched peptides, which feature peptide chains extending from the side chains of internal residues, require coupling chemistries beyond what linear SPPS accommodates. Mirror-image proteins built from D-amino acids are essentially inaccessible through biological expression and technically demanding to produce chemically at useful scale and purity.

Each of these structural classes has strong therapeutic rationale. Cyclic and branched structures offer improved metabolic stability and can engage targets that linear peptides cannot reach. Highly hydrophobic sequences are often required for membrane-associated targets. D-amino acid peptides offer proteolytic stability that L-peptides cannot match.

The synthesis gap, in this dimension, is the distance between structural features that medicinal chemists would like to build into their candidates and structural features that the available manufacturing infrastructure can reliably produce. Amide’s AFPS approach, which uses flow chemistry to generate highly reactive intermediates in situ and maintains excellent heat and mass transfer throughout synthesis, addresses several of these structural complexity challenges simultaneously, making cyclic constructs, branched architectures, and hydrophobic sequences accessible rather than exceptional.

Dimension 3: Sequence-Dependent Synthesis Failures

The third dimension of the synthesis gap is the most difficult to predict and the most damaging to program momentum: sequence-dependent synthesis failures. These are failures that arise not from the general properties of the peptide class but from the specific combination of amino acids in a particular sequence, and they are largely unpredictable from the sequence alone.

In conventional SPPS, certain amino acid combinations create local aggregation, secondary structure formation, or coupling kinetics problems that are not foreseeable from the sequence on paper. A sequence that looks reasonable based on its component residues may fail repeatedly at a specific position in the chain. The failure is reproducible, but diagnosing its cause and finding a workaround requires iterative synthesis attempts, each of which takes time and consumes material.

For drug discovery programs operating under time pressure, sequence-dependent synthesis failures are particularly costly because they are discovered late. A program may spend weeks designing, modelling, and prioritizing a candidate before synthesis is even attempted. When the synthesis fails, the options are to attempt workarounds, redesign the sequence, or abandon the candidate. Each option has real costs in time and program momentum.

Addressing this dimension requires both technical capability, synthesis methods that are robust across a wider range of sequence contexts, and speed, the ability to attempt synthesis, assess results, and iterate quickly enough that a failed synthesis does not halt a program for weeks. The combination of Amide’s synthesis platform with a four-day delivery window is directly relevant to this third dimension: it shortens the feedback loop between design and physical result to a degree that changes how teams can respond to sequence-dependent failures.

How AI Is Widening the Gap

The synthesis gap would be a significant but manageable constraint if the design side of the pipeline were advancing at the same rate as the manufacturing side. It is not. The design side is accelerating.

AI-driven approaches to peptide drug design have moved with unusual speed in recent years. Machine learning models trained on structural and activity data can now propose novel peptide sequences with specific binding characteristics, generate candidate libraries at scale, and optimize lead compounds across multiple dimensions simultaneously. De novo peptide design, which was computationally intractable a decade ago, is now accessible to research programs without specialized computational infrastructure.

The practical consequence of this acceleration is that AI is generating candidate peptides faster than conventional manufacturing can process them. A computational platform that can propose hundreds of candidates for a given target in a week cannot be meaningfully served by a synthesis supply chain where each candidate requires weeks of production time and has meaningful probability of failing outright due to structural complexity or sequence-specific synthesis issues.

The synthesis gap is not widening because manufacturing is getting worse. It is widening because design is getting better, faster.

There is a deeper issue as well. AI-driven design, because it is not constrained by synthetic accessibility in the way that medicinal chemists working from intuition often are, tends to propose candidates that push structural boundaries. An AI model optimizing for binding affinity, metabolic stability, and oral bioavailability simultaneously may converge on cyclic, branched, or otherwise complex structures that score well across those parameters precisely because their structural features contribute to those properties. The resulting candidates are often exactly the ones that fall into the synthesis gap.

Several groups working at the intersection of AI-driven drug design and peptide chemistry have described this as a tractability problem: the candidates that look most promising computationally are frequently the ones that are hardest to make. Resolving that tension requires manufacturing capability that can keep pace with what design is proposing, which is a capability gap that conventional SPPS vendors cannot fill.

Amide’s positioning in this context is direct. The company’s platform was built to synthesize the peptides that existing methods cannot reliably produce, bridging what its technology pages describe as the gap between traditional solid phase peptide synthesis and biological expression. As AI design tools continue to generate more structurally ambitious candidates, that positioning becomes more valuable, not less.

The Therapeutic Cost of the Synthesis Gap

The synthesis gap has costs that extend well beyond the operational inconvenience of delayed material delivery. At the program level, at the portfolio level, and at the level of what therapeutic opportunities the industry is actually able to pursue, the gap shapes outcomes in ways that are rarely made explicit.

Program-Level Cost: Momentum and Iteration Speed

Drug discovery programs are built around design-build-test cycles. A hypothesis is formulated, a compound is synthesized, and it is tested against a biological system. The result informs the next hypothesis. The faster this cycle runs, the more hypotheses a program can test in a given window of time, and the more quickly it can converge on a lead compound worth advancing.

When synthesis is the rate-limiting step, and when synthesis failures are unpredictable and time-consuming to diagnose, the design-build-test cycle stretches. Programs that might complete five or six hypothesis cycles in a quarter complete one or two. The compounding effect over the life of a drug discovery program is substantial.

The cost is not only in absolute time. It is in the quality of the decisions made. When synthesis turnaround is measured in weeks, teams tend to be more conservative in which candidates they attempt, prioritizing sequences that are more likely to work over sequences that are more therapeutically interesting. The synthesis gap, in this way, does not just slow programs down; it systematically biases them toward the more manufacturable and away from the more promising.

Portfolio-Level Cost: Avoidance of Structurally Complex Candidates

At the portfolio level, the synthesis gap has an effect that is harder to measure but arguably more significant: it discourages exploration of the structural space where many of the most interesting new therapeutic opportunities live.

Cyclic peptides, mirror-image proteins, branched architectures, extended sequences designed to engage large protein-protein interaction interfaces: these are precisely the structural classes that offer differentiated mechanisms of action and address targets that small molecules and conventional biologics cannot reach. They are also the classes most affected by the synthesis gap.

When these candidates are consistently difficult to access, programs learn not to prioritize them. The selection pressure runs against structural complexity and toward structural simplicity, not because the simpler candidates are better therapeutics, but because they are easier to make. The synthesis gap, at this level, is distorting what the peptide drug discovery field collectively pursues.

The Market Context

The peptide therapeutics market, projected to reach $82.6 billion by 2032, is driven partly by the same structural factors that create the synthesis gap. The success of GLP-1 receptor agonists including Ozempic, Mounjaro, and related compounds, as well as established peptide drugs for cancer, osteoporosis, and hormone disorders, has validated the commercial and clinical potential of peptide-based medicines. That validation is drawing more programs into peptide therapeutic development, which means more programs running directly into the constraints the synthesis gap creates.

The demand for complex peptide manufacturing capability is not hypothetical. It is growing, and the supply of that capability has not grown at a comparable rate.

What Closing the Synthesis Gap Requires, Technically

A precise understanding of what the synthesis gap is makes clear what closing it requires. It is not a matter of incremental improvement to existing methods. It is a matter of a different synthetic approach.

Why Incremental SPPS Improvement Has Limits

Conventional SPPS improvement efforts have largely focused on optimizing reagent selection, resin chemistry, and coupling protocols. These optimizations can push the reliable length ceiling modestly upward and can improve yields on moderately difficult sequences. They do not resolve the fundamental coupling efficiency problem at extended length, and they do not address the heat and mass transfer limitations that cause failures on structurally complex targets.

The ceiling is not an optimization problem. It is a consequence of the batch-mode, room-temperature conditions under which conventional SPPS operates. Within those conditions, the ceiling cannot be moved much further than it already has been.

What Flow Chemistry Changes

The key insight behind Amide’s AFPS platform is that flow chemistry fundamentally changes the conditions under which synthesis occurs. Rather than cycling reagents over a static resin bed in batch mode, flow-based synthesis continuously moves reagents through the resin under controlled conditions of temperature and flow rate. This enables several things that conventional batch SPPS cannot achieve.

• Highly reactive intermediates can be generated in situ and delivered to the coupling site before they decompose, improving coupling efficiency at each step.
• Heat and mass transfer can be maintained precisely throughout the synthesis, even as the growing peptide chain changes the physical character of the resin environment.
• The combination of heat and flow enables synthesis conditions that break up secondary structures and aggregates that form during conventional synthesis of long or complex sequences.
• The speed of individual coupling cycles is dramatically reduced, enabling faster synthesis without sacrificing quality.

The result is a platform that can reliably produce peptides up to 120 amino acids in length, access structurally complex targets including cyclic constructs, branched peptides, highly hydrophobic sequences, and mirror-image proteins, and deliver material in four days rather than weeks.

The Role of Robotic Automation and AI

Amide’s platform combines AFPS with robotic automation and AI-assisted process optimization. Robotic automation removes the variability introduced by manual handling in complex synthesis workflows, improving reproducibility across batches and enabling the kind of high-throughput production that drug discovery programs increasingly require. AI integration supports process optimization, allowing the platform to adapt synthesis conditions based on sequence characteristics and prior synthesis outcomes.

This combination, flow chemistry enabling what batch SPPS cannot reach, automation ensuring reproducibility, and AI optimizing parameters across the synthesis workflow, is what the company describes as a third approach to bridging the synthesis gap, distinct from both conventional SPPS and biological expression, and capable of serving the candidate space that neither can reach.

The Synthesis Gap as a Strategic Lens for Drug Discovery Programs

For program leaders and portfolio managers in peptide drug discovery, the synthesis gap is worth treating not just as an operational constraint but as a strategic lens: a way of identifying where capability gaps are shaping decisions that should be shaped by science.

Auditing for Synthesis-Driven Candidate Avoidance

The first application of this lens is retrospective. Most drug discovery organizations have a history of candidate prioritization decisions. Some of those decisions were made on scientific grounds: one candidate had better selectivity, another had a cleaner safety profile. Some, though rarely documented as such, were made on synthesis grounds: one candidate was accessible, another was not.

An honest audit of prior candidate prioritization will often surface patterns of synthesis-driven avoidance. Cyclic candidates consistently deprioritized relative to linear ones. Extended sequences consistently dropped in favor of shorter analogs. Structurally complex leads consistently replaced by simpler alternatives with less compelling profiles. These patterns are worth identifying explicitly, because they represent a set of therapeutic opportunities that were shaped by manufacturing constraint rather than scientific merit.

Rethinking What Is Designable

The second application is prospective. If synthesis capability has expanded, the design parameters that were implicitly constrained by synthesis accessibility can be relaxed.

Computational and medicinal chemistry teams working under the assumption that synthesis will cap out at 40 amino acids and will fail on complex structural features may not be designing to the full range of what is now accessible. When the synthesis ceiling is 120 amino acids and when cyclic, branched, and D-amino acid structures are reliably available, the design space expands substantially. Programs that understand their current synthesis capability can design into that expanded space rather than around the constraints of an older infrastructure.

Accelerating the Design-Build-Test Cycle

The third application is operational. When synthesis turnaround is measured in days rather than weeks, and when the success rate on complex targets is high rather than uncertain, the design-build-test cycle can be run fundamentally differently.

Programs can test more hypotheses per unit time. They can attempt structurally ambitious candidates that would previously have represented too large a time investment if synthesis failed. They can respond to in vitro results with rapid analog synthesis rather than pausing for weeks while new material is made. The compounding effect on program velocity is substantial, particularly in early-stage discovery where the value of each iteration cycle is highest.

Amide’s manufacturing platform was built around this operational reality. The four-day delivery window and 99% on-time delivery rate are not simply service metrics; they are expressions of a philosophy that synthesis speed and synthesis capability should not be treated as competing constraints. The question Amide’s founding team set out to answer, drawn from the Pentelute and Jensen Labs at MIT, was precisely whether chemical peptide synthesis could approach what the ribosome does: fast, precise, and reliable across the full range of what a peptide can be.

Conclusion

The synthesis gap is not a temporary inconvenience that will resolve itself as conventional methods improve. It is a structural feature of the current peptide drug discovery landscape, produced by the divergence between an accelerating design capability and a manufacturing infrastructure that was built for a different era of the field.

It has costs that are real but rarely made explicit: programs slowed by synthesis delays, candidate decisions shaped by manufacturing accessibility rather than scientific merit, and structural classes left underexplored because the supply chain could not support them. As AI-driven design tools continue to generate more structurally ambitious candidates at greater speed, those costs will compound.

Closing the gap requires manufacturing capability that can match what design is proposing: reliable synthesis across extended length, structural complexity, and difficult sequence contexts, with turnaround times that support rather than constrain the design-build-test cycle. That is the problem Amide Technologies was built to solve, and the reason that solving it matters beyond any single program.

For drug discovery organizations that are serious about peptide therapeutics, the synthesis gap is worth treating as a first-order strategic question. Where is it shaping candidate decisions? Where is it slowing programs that should be moving faster? Where are structurally complex candidates being left on the table not because the science does not support them, but because nobody could make them?

Amide Technologies is a Cambridge-based biotech offering on-demand complex peptide manufacturing at unprecedented speed. Developed out of MIT’s Pentelute Lab, Amide’s AFPS platform provides reliable synthesis of peptides up to 120 amino acids, cyclic constructs, branched peptides, mirror-image proteins, and highly hydrophobic sequences, with delivery in as few as four days. To learn more or discuss your program’s synthesis needs, visit amidetech.com or contact the team directly.

Amide Technologies | Cambridge, MA | amidetech.com | Developed at MIT’s Pentelute Lab