The AI claim and its evidence gap
The most ambitious technology story in this week's oncology material is also the least proven. Puja Sapra of AstraZeneca describes a 'lab of the future' in which AI makes predictions, robots run experiments and instruments generate data. In her words: "Where a self-driving car uses sensors and models to navigate its environment, this system uses AI to make predictions, robotic systems to execute experiments, and instruments to generate data."1
Readers should know how far to trust that source. Via News's measured reliability check on the MIT Technology Review piece it comes from, 'How AI helps scientists design the next generation of medicines', found 0% of 11 checked claims held up.1 That does not mean the statements are false. It means our system could not tie them to checkable evidence. They are best read as a company's account of its own strategy: forecasts and opinion, not demonstrated results.
What AstraZeneca says it is building
The claims are specific enough to be tested later. Sapra says proprietary data is the edge: "Data is our differentiator. We've built an intentionally diverse portfolio across multiple disease areas and drug types. All of that data empowers us to fine-tune frontier AI models with richer, more representative training sets."1
She also describes AI models that could "help identify which two or three targets to prioritize based on the underlying biology, then optimize across multiple parameters to balance a molecule's potency, stability, manufacturability, and safety."1 That is a description of multi-specific biologics design, the same family as the PD-1/VEGF bispecifics now moving through the clinic. On automation she says: "This will generate AI-ready data at a scale that traditional workflows cannot match."1
She is candid about the hard part: "One of the hardest problems in de novo design is predicting whether a computationally generated molecule will be safe in the human body."1 Her proposed answer is virtual clinical trials on advanced cell systems and micro-scale organ models. She says these have "the potential to generate enhanced biological signals without traditional testing bottlenecks."1 The words 'potential' and 'could' carry the weight throughout. No approved drug is cited as a product of this loop.
On human involvement, she says: "With more autonomous systems, human oversight remains at the heart of this approach—ensuring explainable and ethical AI for the benefit of patients."1 And: "Scientists will work hand-in-hand with these model systems."1 For a technology audience, the useful reading is that the near-term product is AI-assisted design with people still testing the output.
The molecules on the clock today are not AI-designed
The approvals and filings in this week's dossier are conventional development milestones. Via News's reading is that regulators and capital are rewarding new drug types: bispecifics, personalized mRNA vaccines and protein degraders.2 The dossier does not say these were designed by AI.
The FDA has set a November 14, 2026 goal date for Summit's Biologics License Application for ivonescimab, based on HARMONi trial results.3 The FDA also approved Tecentriq and Tecentriq Hybreza with a fluoropyrimidine and oxaliplatin for adjuvant stage III dMMR colon cancer, based on the Phase III ATOMIC study, as the twelfth U.S. indication for Tecentriq.4 Moderna and Merck plan to present detailed intismeran results at ESMO on October 24 and to engage regulators about potential filings.5
A talent signal points to where the mRNA vaccine race is heading. Everest Medicines hired two former Moderna oncology scientists in one day, October 9: Brandon Coder, formerly Research & Early Translational Lead, Cancer Vaccines, and Lakshmi Srinivasan, formerly Head of Oncology Translational Medicine.67
The checked numbers: what the bispecific bet costs
Summit is the clearest case where checked financial data sits next to a pending decision. Its SEC filings, verified in the Via News substrate, show a company with no revenue: $0 in both FY 2023 and FY 2024.8 Over the same period the losses grew sharply. Net loss was $614.9 million in FY 2023, $221.3 million in FY 2024 and $1.08 billion in FY 2025.8 The FY 2025 loss is nearly five times the prior year's, and per-share losses went from $0.99 to $0.31 to $1.44.8 The dossier does not break down what drove the swings, so we do not speculate.
Cash burned by operations is a steadier measure. It rose from $76.8 million in FY 2023 to $142.1 million in FY 2024 and $322.9 million in FY 2025.8 Cash on hand swung widely. It was $71.4 million at FY 2023 year-end, $104.9 million at FY 2024 and $225.3 million at FY 2025.8 It fell to $106.5 million in Q1 2026 and stood at $419.4 million in Q2 2026.8 Capital spending on equipment and facilities stayed tiny: $128,000 in FY 2023, $139,000 in FY 2024 and $657,000 in FY 2025.8 Whatever the money is buying, it is not factories.
For an ordinary investor, the point is that a company with no sales and a growing cash burn is pricing in a single regulatory outcome. The dossier names Summit's partnerships with big pharma, and AstraZeneca and Daiichi Sankyo appear in the same narrative.2 It does not describe their terms, so we cannot say how much of the burn they cover.
Where the money is already being made
Commercial execution looks stronger on the delivery side. Novocure reported second-quarter 2026 net revenues of $184 million, up 16% year over year, driven by 18% global active patient growth.9 Chief executive Frank Leonard said: "This was our strongest quarter to date, with record net revenues and active patients on therapy."9 It also received a CE Mark for Optune Pax in locally advanced pancreatic cancer, with Germany the first EU market to launch.9
In Europe, the CHMP recommended Gilead's Trodelvy plus Keytruda in PD-(L)1-positive first-line metastatic triple-negative breast cancer.10 Evandro de Azambuja said the opinion "is welcome news for our patients and the entire clinical community."10 Real-world evidence is also accumulating for older targeted drugs: the final MINT analysis of 109 patients across 18 German centres found a median time to next treatment of 14.9 months and a 62.4% best overall response rate for mogamulizumab.11
Our narrative analysis also flags capital moving into care delivery, with McKesson and CD&R acquiring Option Care Health after the earlier Precision Medicine Group deal.2 The dossier gives no deal terms, so we offer no valuation.
What to watch
- October 24: the detailed intismeran data at ESMO, and any word on regulatory filings.5
- November 14: the FDA goal date for ivonescimab. A decision is the first test of whether Summit's rising burn pays off.3
- Demonstrated AI results: the claim to hold AstraZeneca to is an AI-designed molecule reaching the clinic. Until one does, the 'lab of the future' remains a forecast from a source that failed our reliability check.1