Abraham Heifets built Atomwise into one of AI drug discovery's early success stories, raising more than $170 million and landing partnerships the company valued at $5 billion. Now he's betting his next act on a premise that sounded implausible five years ago: turning injectable biologics into pills you can take with breakfast.
His new venture, WonderTx, emerged from Y Combinator's Summer 2026 batch and an audacious pitch. The San Francisco startup says it's building "extrapolative frontier models" capable of designing oral molecules against drug targets that have never been successfully addressed with pills before. Think insulin. Think Ozempic. Think hundreds of billions in annual revenue currently delivered by needle.
The timing matters because the industry just proved it's possible. On April 1, 2026, Eli Lilly won FDA approval for Foundayo, an oral GLP-1 pill with none of the fussy dosing restrictions that plagued earlier attempts. You can take it with food, with coffee, whenever. Three months later, Merck cleared regulatory approval for Lipfendra, the first oral drug to target PCSK9, a cholesterol pathway previously addressed only by injections. Novo Nordisk had already launched an oral semaglutide for obesity in February 2025.
Suddenly a category shift that pharmaceutical executives once dismissed as technically remote looks more like an inevitability.
Goldman Sachs raised its 2030 obesity drug forecast to $114 billion in July 2026, projecting that oral formulations would capture roughly 40 percent of that market within four years. J.P. Morgan pegged the combined diabetes and obesity incretin market at around $200 billion by decade's end. Those projections predate the Lilly and Merck approvals, which suggests analysts may still be underestimating the velocity of change.
Three Paths to Oral Delivery
The pharmaceutical industry didn't crack this problem with a single breakthrough. Companies took divergent technical routes, and the differences matter for what comes next.
Novo Nordisk spent $1.8 billion in 2020 to acquire Emisphere Technologies, gaining access to SNAC, a permeation-enhancer molecule that helps peptides survive the digestive tract. That technology enabled the first oral semaglutide approvals, though the resulting product, Rybelsus, still requires patients to take it first thing in the morning on an empty stomach with no more than four ounces of water and then wait 30 minutes before eating or drinking anything else. The FDA label runs to several pages of dosing instructions.
Lilly avoided that complexity by designing orforglipron as a small-molecule GLP-1 agonist from the ground up rather than trying to reformulate an existing peptide. The company said the drug can be taken any time of day without food or water restrictions. It's a fundamentally different molecule that happens to hit the same biological target.
Merck's enlicitide represents yet another modality: an oral macrocyclic peptide, a ring-shaped molecule larger than traditional small-molecule drugs but smaller than full proteins. The FDA approved it for adults with high cholesterol in July 2026, marking the first time a PCSK9 inhibitor could be taken as a pill. Previously that pathway was the exclusive domain of injectable monoclonal antibodies.
Other oral alternatives have quietly reached the market over the past few years. BioCryst won FDA approval in December 2020 for Orladeyo, an oral treatment for hereditary angioedema that replaced injectable therapies, then added a pediatric pellet formulation last December. Chiasma's Mycapssa became the first oral somatostatin analog in June 2020.
Not every program succeeded. Pfizer discontinued danuglipron, its oral GLP-1 candidate for chronic weight management, in December 2025 after safety concerns surfaced, including a potential case of drug-induced liver injury. Oramed's oral insulin candidate failed Phase 3 endpoints in January 2023, prompting the company to seek partners for protocol redesign.
The setbacks underscore the technical difficulty. Biologics evolved as injectables partly because proteins and peptides degrade rapidly in stomach acid and struggle to cross the intestinal wall into the bloodstream. Solving that problem requires either chemical modification that preserves activity, new delivery mechanisms, or designing entirely new molecules.
Device Platforms Enter the Race
A parallel group of companies is attacking oral delivery not by redesigning drugs but by building platforms that force existing biologics through the intestinal barrier.
Rani Therapeutics developed the RaniPill, a robotic capsule that travels through the digestive system and injects drug payload transmucosally in the GI tract. The company reported preclinical data on oral semaglutide delivered via the device in February 2025 and announced a partnership with ProGen for obesity therapy in June 2024.
i2O Therapeutics raised seed funding in April 2020 from Sanofi Ventures and the JDRF T1D Fund for its ionic-liquid platform designed to carry oral macromolecules across the gut wall. The startup later signed collaborations with both Sanofi and Janssen, according to Harvard's Office of Technology Development.
Entera Bio partnered with OPKO in 2025 on an oral GLP-1/glucagon dual agonist using its N-Tab technology, then expanded a separate collaboration with Amgen in February 2026 to include an oral parathyroid hormone for hypoparathyroidism. Enteris BioPharma, acquired by SWK in 2019, had licensed its Peptelligence platform to Ferring back in 2015, SEC documents show.
MIT researchers and Novo Nordisk published foundational work on SOMA and L-SOMA ingestible self-orienting injectors in Science in 2019. The devices mechanically position themselves in the stomach and deploy microneedles to inject biologics through the gastric wall, bypassing digestive enzymes entirely. Follow-on studies continued through 2022.
Whether device-based approaches can scale to commercial manufacturing and gain patient acceptance remains unclear. Swallowing a pill feels different from swallowing a robotic capsule, even if both avoid needles.
Payer Pushback Reshapes Coverage
The economics of GLP-1 drugs have forced employers and insurers into uncomfortable positions. A February–March 2026 survey by the Business Group on Health found that eight in 10 employers identified GLP-1 medications as drivers of higher costs, with many reconsidering coverage policies.
A Becker's analysis in July 2026 found that 36 percent of employers cover GLP-1s for both weight loss and diabetes, 60 percent cover diabetes only, and 3 percent offer no coverage at all. Managed Healthcare Executive reported in September 2026 that roughly 65 percent of a PBM purchasing group's clients excluded GLP-1s for weight loss in the first half of this year.
Medicare's GLP-1 Bridge program went live July 1, 2026, establishing prior-authorization criteria for coverage across type 2 diabetes, obstructive sleep apnea, fatty liver disease, and several other conditions. The FDA's March 2024 label expansion for Wegovy to include cardiovascular risk reduction had opened the door to Part D coverage for that indication.
Oral formulations may ease some payer resistance if they prove cheaper to manufacture and distribute than injectables, though bioavailability constraints complicate that calculation. Low oral bioavailability means higher API requirements per dose, which can offset savings from eliminating cold-chain logistics and injection devices.
AbbVie's experience with oral CGRP migraine drugs offers a preview. The company reported combined 2025 net revenues of $2.307 billion for Qulipta and Ubrelvy, reflecting continued growth as adoption rises against injectable CGRP antibodies. Patients and providers gravitated toward pills despite the injectables' superior efficacy in clinical trials.
The Patent Cliff Accelerates
IQVIA forecast that 118 biologics representing $232 billion in U.S. sales will lose exclusivity between 2025 and 2034, with the steepest drop-off hitting in 2028. Evaluate and IQVIA separately outlined more than $300 billion in biologic revenue at risk by 2030 as patents expire.
For pharmaceutical companies facing that cliff, oral alternatives offer a path to extend franchises. A fundamentally different molecule can secure new composition-of-matter patents even if it targets the same pathway as an expiring biologic. Lilly acquired DICE Therapeutics for approximately $2.4 billion in mid-2023 to gain oral IL-17 antagonists discovered via DNA-encoded libraries. Sanofi advanced balinatunfib, an oral small-molecule inhibitor of TNF-TNFR1 signaling, into Phase 2 trials in 2024–2025, though the company's future strategy for the program remained under evaluation as of October 2025.
WonderTx cited the DICE acquisition in an August 2026 blog post as evidence that oral drugs can address protein-protein interactions historically labeled "undruggable." The company said it's building "a lights-out wonder drug factory, where AI runs the entire design-make-test loop."
Heifets assembled a technical team with unusual depth. Saulo H. P. de Oliveira, one of the co-founders, authored tools including qFit3 and KVFinder and worked on synthesizability in generative chemistry AI, the company said. Cameron Ferroni, the third co-founder, co-founded Xbox and Xbox Live at Microsoft before serving as CTO at PhenoTips in 2025.
WonderTx said it has achieved internal validation on four targets without on-target training data, a claim that points to the company's stated focus on "zero-shot drug discovery" against novel targets. Dealroom listed a $125,000 seed round from Y Combinator in September 2026, the standard investment for the accelerator, though the company declined to disclose additional funding details.
The company's pitch centers on pursuing targets already validated in humans through existing antibodies or peptide therapies, which it argues reduces translational risk. Whether AI models can reliably design oral molecules against targets where no oral drugs currently exist remains an open empirical question. The field lacks enough published data to assess success rates independent of company claims.
Approved oral biologics now span GLP-1, PCSK9, hereditary angioedema, somatostatin analogs, and CGRP pathways. Mid-stage programs are advancing in TNF and IL-17. Goldman Sachs and Morgan Stanley both flagged manufacturing intensity as a consideration, given that low-bioavailability oral peptides require substantially more active pharmaceutical ingredient than injectables delivering the same therapeutic dose.
The shift from injectable to oral formulations is no longer speculative. It's underway, unevenly distributed across disease areas but moving faster than most forecasts anticipated just two years ago. Whether AI-designed molecules can accelerate that transition or simply join a crowded field of platform plays will depend on clinical data that doesn't yet exist.
For now, WonderTx is placing a bet that the same computational approaches that worked in virtual screening can extend to designing entirely new chemical matter. Heifets built one AI drug discovery company through a decade of hype cycles and market skepticism. This time the market already believes the destination is real. The question is who gets there first, and by what route.
