KRAS was called undruggable for forty years, and this week it took another beating. Genentech’s divarasib showed superiority over approved therapy in a Phase 3 trial in previously treated non small cell lung cancer. That matters for two reasons. First, head to head wins are rare in oncology. Most drugs get approved by beating chemo or placebo, not by beating the current targeted standard, so a G12C inhibitor topping approved therapy is a real statement. Second, it reshapes a competitive field that Amgen’s Lumakras opened in 2021 and that BMS and Mirati’s Krazati joined soon after. Both of those drugs underwhelmed commercially relative to the early hype, and divarasib now arrives claiming it can do better.
Step back and the whole RAS story is the one to watch this year. We spent the spring on Revolution Medicines and its pan RAS drug daraxonrasib, which posted practice changing pancreatic cancer data at ASCO, and its second molecule RM 055. Now Genentech is pushing the narrower G12C approach past the incumbents. Two different bets on the same oncogene, both paying off in the same year, after four decades of failure. That is the kind of arc that defines a decade in cancer drug development.
Separately, Novartis grabbed another deal, paying up to $1.5 billion for Myricx Bio and its novel ADC payload platform. And Anthropic CEO Dario Amodei walked back some of his own bold predictions about AI transforming drug discovery.
Genentech Made the Case That It Has the Best KRAS G12C Drug
What Happened: Roche’s Genentech unit reported that its KRAS G12C inhibitor divarasib beat approved therapy in a Phase 3 trial in previously treated non small cell lung cancer, according to BioWorld.
What a Head to Head Win Actually Means
Most oncology drugs earn their approval by demonstrating superiority over chemotherapy, placebo, or best supportive care. That is a lower bar than beating an approved targeted therapy. When a drug wins head to head against the current standard—in this case, the approved KRAS G12C inhibitors that represent the best available targeted treatment for this mutation—it makes a stronger claim to being the drug physicians should prescribe first.
The G12C mutation drives a meaningful slice of lung cancers. Amgen’s Lumakras arrived in 2021 as the first drug to hit this target and was celebrated as a breakthrough after decades of failure. Krazati from Mirati (now part of BMS) followed. Both were scientific milestones. Neither became the commercial blockbuster the early excitement suggested. Response durations were modest. The drugs worked, but not as long or as decisively as the market hoped.
Divarasib now enters claiming it can do better. If the full data hold up—and we will need to see the full dataset at an upcoming medical meeting to judge the size of the benefit—Genentech has the best in class G12C drug and the rest of the field has a problem. A marginal edge changes prescribing gradually. A decisive one changes it fast. The magnitude matters.
The Two Track RAS Story
The divarasib win connects to the bigger RAS narrative we have tracked all year. Two companies are placing two different bets on the same family of oncogenes, and both are paying off simultaneously.
Genentech is going narrow and deep on G12C, the single most common druggable KRAS mutation. G12C is largely a lung cancer story today, and divarasib’s Phase 3 win positions it at the top of that specific market. The focus is precise: one mutation, one tumor type, best in class.
Revolution Medicines is going broad with daraxonrasib, a pan RAS approach that hits across RAS mutation subtypes. Daraxonrasib posted practice changing data at ASCO (OS HR 0.40 in second line pancreatic cancer, NEJM published, expanded access granted, CNPV filing imminent). Revolution’s second molecule, RM 055, doubled the chemotherapy response rate in pancreatic cancer through a different mechanism. The pan RAS approach opens pancreatic and colorectal cancers where the mutations are more varied than the single G12C alteration that dominates in lung cancer.
These are not competing for the same patients. G12C inhibitors serve the subset of lung cancer patients with that specific mutation. Pan RAS inhibitors serve a broader population across multiple tumor types. But they are both proving that RAS, the target the field gave up on for four decades, is now one of the most productive areas in oncology. Two bets, two wins, one year. After forty years of failure, the field has an embarrassment of riches.
Novartis Pays Up to $1.5B for a Better ADC Payload
What Happened: Novartis agreed to acquire UK based Myricx Bio for up to $1.5 billion ($1.1 billion upfront plus up to $400 million in milestones), adding a next generation ADC payload platform built around a novel NMT inhibitor mechanism.
Why the Payload Matters Now
Most ADC dealmaking this year has focused on targets—the antibody’s address label that tells the drug which cells to find. HER2, Trop 2, B7 H3, Nectin 4, and others. Companies have been racing to stake claims on the best targets. Novartis made a different bet. The value in the Myricx deal is not the target. It is the payload—the toxic warhead the antibody delivers once it arrives at the tumor.
That distinction matters because the payload has quietly become the bottleneck in ADC development. Current ADC payloads mostly fall into a small number of chemical classes. As more ADCs reach patients and more patients receive them, two problems are emerging. First, tumors develop resistance to the existing payload classes over time, the same way they develop resistance to any drug. Second, when patients receive multiple ADCs sequentially (which is increasingly common in oncology treatment algorithms), the overlapping toxicity from similar payloads limits what physicians can give safely.
Myricx’s platform is built on NMT inhibition, a distinct mechanism of killing cancer cells that could work where existing payloads fail. Rather than using the same chemical warheads that every other ADC deploys, the NMT approach offers a fundamentally different way to destroy the tumor once the antibody delivers its cargo. Pairing a novel payload with established targets like B7 H3 and HER2 is a way to refresh the entire ADC toolkit rather than chase one more target.
This is a smart read of where the ADC field is heading. After years of racing to new targets, the technology frontier is shifting to the payload itself. Whoever controls a differentiated, resistance breaking payload has leverage across many targets at once—the same novel warhead can be attached to antibodies aimed at different tumor antigens. That platform logic is what makes a preclinical to early stage platform worth $1.5 billion to Novartis, and it is the same logic that will drive other large ADC players to go looking for their own payload edge.
The Loudest Voice in AI Just Told Everyone to Slow Down
What Happened: Anthropic CEO Dario Amodei tempered his own earlier predictions about how quickly AI will transform drug discovery, according to STAT.
Why This Reset Is Good for the Field
We have covered a steady run of AI discovery deals this year. Isomorphic Labs ($2.7 billion in total funding) working with Lilly and Novartis on protein design. Insilico Medicine partnerships with SK Biopharmaceuticals ($2.5 billion), Servier ($888 million), and Takeda. The Anthropic pharma ecosystem spanning five of the top ten global pharmaceutical companies. Alnylam and Inceptive optimizing RNA sequences. Lilly and Profluent ($2.25 billion) designing AI recombinases.
The technology is real. AI is already useful in target identification, molecular design, and sequence optimization. The deals are real. The capital is flowing at institutional scale.
What was never realistic was the compressed timeline. The idea that AI would collapse a decade of drug development into a couple of years was always more pitch than plan. Biology does not speed up just because the computational chemistry does. You still have to run the clinical trials. Trials take years. Trials fail often. The most powerful AI platform in the world cannot make a Phase 3 trial enroll faster or make human biology less complex.
When the CEO of one of the leading AI companies says this out loud, it recalibrates expectations in a useful direction. Realistic timelines make for better partnerships and fewer disappointed boards. The companies that will capture the most value from AI in drug development are the ones using it to find better candidates faster—not the ones promising that AI will eliminate the clinical risk that has defined this industry for decades.
Our view: this is healthy. A little realism from the top is better for the field than another round of overpromising. The tools work. The revolution will just take longer than the pitch deck said.
Medicare Reopened the 340B Fight
What Happened: Medicare proposed steep cuts to 340B drug payments, according to STAT, reigniting one of the most contentious ongoing fights between hospitals and the government.
The 340B program requires drug manufacturers to sell outpatient drugs at discounted prices to hospitals and clinics that serve low income patients. The program has ballooned in size and complexity over the past decade, and critics argue that some participating hospitals use the discounts to generate margin rather than to expand care for vulnerable populations. The administration’s proposed cuts target the payments Medicare makes for drugs purchased through the 340B channel.
Expect a loud comment period. The hospital lobby is one of the most effective in Washington, and any move to cut 340B payments draws immediate, organized pushback. For drug manufacturers, the program remains a persistent source of channel complexity and disputed discounts. Hospitals buy drugs at 340B prices, sometimes administer them to patients with commercial insurance, and pocket the difference between the discounted acquisition cost and the higher commercial reimbursement. Manufacturers have pushed back through contract pharmacy restrictions and data transparency demands—Lilly set a 340B compliance deadline earlier this month.
Any reform that clarifies the rules could cut both ways. Stricter oversight might reduce the program’s scope (bad for hospitals, good for manufacturers who feel the discounts are being exploited). But more aggressive payment cuts could also compress the pricing that manufacturers receive in the 340B channel, adding another layer of margin pressure on top of the MFN framework, Section 232 tariffs, and Medicaid rebates.
Venture Capital Keeps Flowing at Scale
One more data point on the health of the capital markets: nearly 60% of venture rounds in the first half of 2026 topped $100 million, with those large deals collectively worth $7.5 billion—up 23% from a year ago.
That number tells you something important about where the private biotech market stands. The venture environment is not just open. It is tilted heavily toward large, concentrated bets on later stage companies with validated science. Smaller seed and Series A rounds are still happening, but the capital gravity is pulling toward the $100 million plus raises that fund pivotal trials, commercial preparation, and the kind of clinical programs that attract acquisition interest from large pharma.
Combined with the 13 IPOs raising $4.1 billion, the $805 million Definium offering, the $1 billion BridgeBio preferred equity raise, and the $1.3 billion Apogee/Blackstone deal (before AbbVie acquired it for $10.9 billion), the capital picture is the strongest it has been since 2021 at every level: venture, IPO, follow on, and structured financing.
Strategic Themes
1. KRAS Now Has a Best in Class Fight, and the Whole RAS Space Is on Fire
Genentech’s divarasib claims superiority in G12C lung cancer. Revolution’s daraxonrasib doubled survival in pan RAS pancreatic cancer. RM 055 doubled the chemo response through yet another mechanism. AbbVie and Kestrel have a pan KRAS inhibitor in development. The target that was undruggable for forty years now has multiple drugs, multiple mechanisms, and multiple companies competing across multiple tumor types. This is what it looks like when an entire therapeutic class comes of age in a single year.
2. Novartis Recognized the ADC Bottleneck Before Most of the Field
Most ADC deals in 2026 have been about targets. Novartis paid $1.5 billion for a payload. That difference matters because the next wave of ADC competition will not be won by finding new targets alone—it will be won by whoever has the most potent, most durable, most resistance proof warhead to attach to those targets. The NMT inhibitor platform from Myricx could become Novartis’s competitive moat across the entire ADC modality, the same way Novartis’s $7.1 billion molecular glue investment is designed to own the protein degradation space.
3. Amodei’s AI Reality Check Does Not Diminish the Technology—It Protects It
When the hype cycle outpaces the technology, disappointment is inevitable. Amodei tempering his own predictions is the best thing that could happen for AI drug discovery right now, because it shifts the conversation from “AI will revolutionize everything tomorrow” to “AI is genuinely useful for specific parts of the process and the returns will come over years, not months.” That framing attracts the kind of long term capital and partnership commitment that the field actually needs, rather than the short term enthusiasm that leads to deal remorse.
4. The 340B Fight Will Get Loud But the Outcome Affects Every Drug’s Channel Economics
The 340B program touches every manufacturer, every covered entity hospital, and every patient who receives outpatient drugs through a participating institution. Proposed cuts to 340B payments do not directly change the price a manufacturer receives, but they reshape the economics of the channel in ways that affect volume, mix, and margin. With MFN pricing, Section 232 tariffs, Medicare Bridge copays, and European pricing reforms all running simultaneously, the 340B proposal adds another variable to an already complex pricing environment.
Frequently Asked Questions
What did divarasib show?
Genentech’s KRAS G12C inhibitor beat approved therapy in a Phase 3 trial in previously treated NSCLC. Head to head wins over approved targeted therapies are rare in oncology. It positions divarasib as a potential best in class G12C drug ahead of Amgen’s Lumakras and BMS/Mirati’s Krazati, both of which underwhelmed commercially.
How does this fit with Revolution Medicines?
Two different bets on the same oncogene family. Genentech is going narrow on G12C (primarily lung cancer). Revolution is going broad with pan RAS inhibition (pancreatic, colorectal, and beyond). They serve different patient populations through different mechanisms. Both are winning in the same year after forty years of KRAS being undruggable.
What is the Novartis/Myricx deal?
Up to $1.5 billion ($1.1 billion upfront) for a novel ADC payload platform built on NMT inhibition. The deal targets the payload bottleneck rather than racing to new ADC targets. Lead programs aim at B7 H3 and HER2 with a differentiated warhead designed to overcome resistance to existing payload classes.
What did Amodei say about AI?
The Anthropic CEO tempered his own predictions about how fast AI will transform drug discovery. The reset acknowledges that while AI genuinely accelerates candidate identification and molecular design, clinical trials still take years and biology remains complex. The technology is real but the timelines were overpromised.
What are the 340B cuts?
Medicare proposed steep cuts to 340B drug payments, which would reduce what Medicare pays for drugs purchased through the 340B discount program by hospitals serving low income patients. Expect significant hospital lobby pushback during the comment period.
How is the venture market?
Nearly 60% of venture rounds in H1 topped $100 million. Those large deals were worth $7.5 billion, up 23% from a year ago. The capital environment continues to favor large, concentrated bets on later stage companies with validated science.
BioMed Nexus Pro — What Institutional Subscribers Are Reading Today
KRAS Finally Has a Best in Class Fight. We break down where divarasib leaves Amgen and BMS commercially, how it fits alongside Revolution’s pan RAS push, and which companies end up owning which slice of the RAS oncology market over the next decade.
Novartis Paid for a Payload, Not a Target. We analyze why the ADC bottleneck is shifting from targets to payloads, what NMT inhibition offers that existing warhead classes do not, and which other large ADC players need to solve the same problem.
Amodei Hit the Brakes on AI Hype. We lay out which AI discovery deals have a credible path to clinical returns and which are running on narrative. The technology works. The timelines do not. Here is how to tell the difference.
Plus: 340B payment reform analysis, venture capital scorecard, July 17 China deadline tracker, Revolution filing watch, and the full H2 catalyst calendar.
About BioMed Nexus
BioMed Nexus delivers institutional grade intelligence to biotech and pharma executives, investors, and clinicians. Our daily briefings and deep dive analyses cut through the noise to deliver the strategic insights that drive better decision making in the life sciences.
Subscribe to receive daily updates and gain access to BioMed Nexus Pro institutional intelligence briefs.
