Highlights & Summary

June brought a run of steady progress across drug discovery, diagnostics, and healthcare delivery, but two moves stood out for a different reason: frontier AI companies pushing directly into life sciences and clinical care. Anthropic launched Claude Science, an AI workbench for research, alongside its own internal drug-discovery program aimed at neglected diseases. And Midjourney, known for consumer image generation, unveiled a healthcare division and a radiation-free full-body scanner. Both are early and unproven, but they signal that the largest AI labs now see healthcare and drug discovery as key areas of focus.

‍

Headlines

AI in Drug Discovery  

1 | Insilico Medicine and SK Biopharmaceuticals form AI-powered drug discovery collaboration worth up to $2.5 billion for neuroimmune disorders | Partnership

2 | Anthropic launches Claude Science and an internal drug discovery program for neglected diseases | New Product

3 | NewLimit raises $435M Series C led by Founders Fund to bring longevity medicines to human trials | Financing

4 | Owkin to build AI agents as part of a multi-year K Pro collaboration with Sanofi | Partnership

5 | XtalPi announces over $400 million partnership with a leading biopharma to develop oral small-molecule therapeutics for a GPCR target | Partnership

‍

‍

‍

AI in Diagnostics  

1 | Midjourney launches Midjourney Medical and a full-body “Ultrasonic CT” scanner | New Product

2 | Pathway Labs announces world’s first FDA clearance for multicondition AI in cardiology and a partnership with OpenEvidence | Regulatory

3 | Aidoc receives FDA Breakthrough Device Designation for AI that drafts radiology reports | Regulatory

4 | Proscia launches Fifth Generation of Concentriq with embedded multimodal AI | New Product

5 | Tempus launches open-source digital pathology consortium with Yale New Haven Health and Memorial Sloan Kettering | New Product

‍

‍

AI in Healthcare  

1 | Abridge unveils AI-native clinician intelligence platform organized around the patient | New Product

2 | UpDoc debuts first FDA-cleared clinical AI platform built on patient-facing LLMs | Regulatory

3 | Google releases MedGemma 1.5 and MedASR for medical imaging and speech-to-text | New Product

4 | Innovaccer signs multi-year strategic collaboration with AWS to scale agentic healthcare AI | Partnership

5 | FDA refreshes its list of AI-enabled medical devices | Regulatory

‍

‍

‍

‍

Summaries

AI in Drug Discovery  

1 | Insilico Medicine and SK Biopharmaceuticals form AI-powered drug discovery collaboration worth up to $2.5 billion for neuroimmune disorders | Partnership

Announced at the BIO 2026 International Convention, Insilico will apply its Pharma.AI platform (spanning target validation, generative chemistry, and molecule optimization) to discover candidates for neuroimmune CNS disorders, while SK Biopharmaceuticals leads late-stage development and U.S. commercialization. Insilico receives up to $18 million in upfront and near-term milestones, with total potential value exceeding $2.5 billion in development, regulatory, and commercial milestones plus single-digit royalties. It is Insilico's largest deal with an APAC partner to date and its second multi-billion-dollar pharma collaboration this year, following an up-to-$2.75 billion agreement with Eli Lilly in March.

2 | Anthropic launches Claude Science and an internal drug discovery program for neglected diseases | New Product

Anthropic introduced Claude Science, an AI workbench that unifies 60+ scientific databases, coding tools, on-demand compute, and reproducibility features (including a reviewer agent that checks citations and calculations) into a single research environment for genomics, proteomics, structural biology, and cheminformatics. Notably, it is not a new model, running on existing Claude models including Opus 4.8, and is available in beta to all paid subscribers. Alongside the launch, Anthropic said it is starting its own preclinical drug discovery program focused on "neglected" diseases that traditional pharma considers commercially unattractive, framing the effort as a way to build credibility with the biopharma customers it hopes to win.

3 | NewLimit raises $435M Series C led by Founders Fund to bring longevity medicines to human trials | Financing

NewLimit, the epigenetic-reprogramming company co-founded by Coinbase CEO Brian Armstrong, closed a $435 million Series C at a roughly $3.1 billion valuation, more than triple its level a year earlier, with Eli Lilly Ventures among the returning backers. The company uses its proprietary AI model and genomics tools to identify transcription-factor combinations that restore youthful function in aged cells, and plans a first-in-human trial of its lead LNP-RNA liver-reprogramming therapy next year in fatty/alcohol-related liver disease. The raise ranks among the largest private biotech financings of the month and reflects investor appetite for platforms targeting the biology of aging itself rather than single indications.

4 | Owkin to build AI agents as part of a multi-year K Pro collaboration with Sanofi | Partnership

Owkin will lead end-to-end development of custom "biopharma agents" for Sanofi, deployed through a five-year license for K Pro, Owkin's agentic AI Scientist that combines multimodal patient data with specialized biological AI across the drug development value chain. The agents are designed to autonomously perform complex R&D tasks and reinforce Sanofi's existing agentic AI capabilities. The deal extends a partnership that began in 2021 as a EUR 90 million collaboration on oncology target identification and patient subgrouping, and signals pharma's shift toward embedded, agentic AI over narrow prediction tools.

5 | XtalPi announces over $400 million partnership with a leading biopharma to develop oral small-molecule therapeutics for a GPCR target | Partnership

Following a successful pilot, XtalPi entered a partnership with an undisclosed international biopharma to design a best-in-class oral small molecule against a difficult metabolic GPCR characterized by high conformational plasticity and no public co-crystal structures. The partner provides an upfront payment and fully funds early R&D, with XtalPi eligible for preclinical, clinical, and commercial milestones plus royalties, bringing total potential value to over $400 million. XtalPi deployed enhanced-sampling simulations and its XFEP free-energy-perturbation platform to screen candidates, and will run design-make-test-analyze cycles through its closed-loop quantum-physics, generative-AI, and robotics engine; it is the largest single-program deal disclosed by the Hong Kong-listed platform to date.

‍

‍

‍

‍

AI in Diagnostics  

1 | Midjourney launches Midjourney Medical and a full-body “Ultrasonic CT” scanner | New Product

The AI image-generation company unveiled a healthcare division and a radiation-free whole-body imaging system that rings the body with roughly 358,000 ultrasound transducers to reconstruct a 3D cross-section in about 60 seconds, using ultrasound-on-chip technology licensed from Butterfly Network. Founder David Holz framed it as the first new whole-body imaging method in decades, with image quality the company claims rivals MRI, though the eye-catching figures come from an early prototype and are not independently verified. It is a Gen-1 device with no FDA diagnostic clearance; Midjourney plans to start with body-composition maps rather than diagnosis, open a first “spa” location in San Francisco in late 2027, and eventually deploy 50,000 scanners.

2 | Pathway Labs announces world’s first FDA clearance for multicondition AI in cardiology and a partnership with OpenEvidence | Regulatory

The FDA cleared EchoNext, the first AI tool to detect hidden structural heart disease from a standard ECG, with the clearance spanning six indications. On June 22, Nature Medicine published a world-first peer-reviewed case in which EchoNext flagged undiagnosed heart failure that had been missed, ultimately leading to a heart transplant. A partnership with OpenEvidence, the clinical decision platform used by over 500,000 physicians, will surface EchoNext's screening output at the point of care, addressing the years-long lag between FDA approval and real-world clinical reach.

3 | Aidoc receives FDA Breakthrough Device Designation for AI that drafts radiology reports | Regulatory

The FDA granted Breakthrough Device Designation to First Read, Aidoc's AI for analyzing chest radiographs and generating preliminary radiology report text, with the designation covering four life-threatening findings. Built on the same architecture as Aidoc's FDA-cleared Triage application and powered by its CARE foundation model and aiOS platform, the tool targets the widening gap between imaging demand and interpretation capacity as turnaround times climb. It is Aidoc's second Breakthrough Device Designation in under a year, following CARE Triage in September 2025, and signals the agency's openness to generative AI that drafts, not just flags.

4 | Proscia launches Fifth Generation of Concentriq with embedded multimodal AI | New Product

Proscia introduced the most significant overhaul of its Concentriq digital pathology platform in five years, built on a new AI-native architecture that brings domain-specific vision, language, and multimodal models into the core to serve both diagnostic labs and biopharma across the drug development lifecycle. The architecture unifies images, metadata, and case or study context so experts and AI models work from the same complete picture, with agentic features accelerating case review, automating lab operations, and supporting biomarker discovery. Concentriq is trusted by diagnostic laboratories managing 12 million patient cases annually and 16 of the top 20 pharmaceutical companies, with early-access deployments beginning in June and broader availability later in 2026.

5 | Tempus launches open-source digital pathology consortium with Yale New Haven Health and Memorial Sloan Kettering | New Product

Tempus, alongside founding academic members Yale New Haven Hospital and Memorial Sloan Kettering, launched an open-source Image Management System (IMS) Consortium to standardize and democratize digital pathology across academic medical centers and industry partners. To seed the effort, Tempus is open-sourcing the existing Paige IMS, including the FullFocus slide viewer, FullFolio case management solution, and AI orchestration and integration modules, providing institutions a clinical-grade baseline platform and viewer. The move positions the market leader to set shared standards and interoperability for a field where fragmented, proprietary systems have slowed adoption.

‍

‍

‍

AI in Healthcare  

1 | Abridge unveils AI-native clinician intelligence platform organized around the patient | New Product

At a June 11 keynote in New York City, ambient-documentation leader Abridge introduced what it calls the first AI-native clinician intelligence platform, expanding well beyond note-taking into a system that coordinates the clinical, financial, and evidence-based decisions across a patient encounter. The platform brings intelligence into workflows clinicians already use, preparing them before a visit, supporting the conversation during it, and reducing administrative burden afterward. The move repositions the category leader, valued at roughly $5.3 billion and deployed across more than 250 health systems, from a documentation tool toward an integrated layer spanning documentation, coding, and clinical decision support.

2 | UpDoc debuts first FDA-cleared clinical AI platform built on patient-facing LLMs | Regulatory

On June 25, Palo Alto-based UpDoc announced FDA clearance for the first Software as a Medical Device that uses patient-facing large language models, advancing an emerging category the company defines as clinical AI, or AI that delivers care historically requiring a licensed clinician. UpDoc positions the platform as designed to support clinicians rather than replace them, arriving as consumer-facing AI doctors reach market with limited regulatory scrutiny. The clearance is paired with initial deployments at four of the nation's leading health systems.

3 | Google releases MedGemma 1.5 and MedASR for medical imaging and speech-to-text | New Product

On June 30, Google Research released its most capable open medical multimodal models in the MedGemma collection alongside MedASR, an open speech-to-text model built for the specialized vocabulary of clinical dictation. MedASR cut transcription errors substantially against the general-purpose Whisper large-v3 model, with 58% fewer errors on chest X-ray dictations and 82% fewer on an internal benchmark spanning diverse specialties and speakers. Because the models are open-weight, developers retain control over privacy and infrastructure, lowering the barrier to building multimodal medical AI tools; early adopters include health authorities in Malaysia and Taiwan.

4 | Innovaccer signs multi-year strategic collaboration with AWS to scale agentic healthcare AI | Partnership

On June 25, healthcare data and analytics company Innovaccer signed a multi-year strategic collaboration agreement with Amazon Web Services to help health systems and payers deploy agentic AI at scale. The collaboration targets the gap between working AI pilots and production deployments that meet security, compliance, and scale requirements, with Innovaccer running agentic workloads on AWS services and expanding distribution through the AWS marketplace. CEO Abhinav Shashank framed the effort around the view that AI in healthcare has a production problem rather than an innovation problem; the agreement also includes co-investment in customer success programs.

5 | FDA refreshes its list of AI-enabled medical devices | Regulatory

On June 16 the FDA updated its public list of AI- and machine-learning-enabled medical devices authorized for marketing, adding devices with final decisions through late March 2026 and reflecting record recent submission volume. Radiology continues to dominate the field by a wide margin, followed by cardiovascular and neurology applications. The agency also signaled it is developing methods to identify and tag devices incorporating foundation models and large language models in future updates, and reiterated that the list is not a comprehensive inventory.