MindRank’s oral GLP-1 is the furthest along any AI-native drug programme
THE SO WHAT
Taking an oral GLP‑1 from program start to Phase III in ~4.5 years on ~$23M is a concrete compression of both time and capital for AI-native drug discovery. Pharma and biotech teams should be recalibrating build-vs-partner decisions — especially for small-molecule programs where AI can shrink early-stage burn dramatically.
READ THE SOURCE
MORE FROM THE WIRE
Deep & Emerging TechHuawei Pitches Egypt on Building AI Data Centers
Huawei proposing AI data centers for Egypt’s military and public sector underscores how AI infra is becoming a geopolitical export, not just a commercial product. Any organization plugging into these stacks needs to weigh long-term data residency, vendor dependence, and alignment with national security policy — not just cost per FLOP.
Deep & Emerging TechSpaceX is spending $100bn on a second Starbase, on an old Exxon site in Louisiana
A $100B self-sustaining spaceport on a former Exxon site is industrial policy by other means—tying heavy launch, energy, and Gulf Coast logistics into a single private node. For operators in energy, petrochem, or heavy manufacturing, expect downstream pressure on Gulf infrastructure, talent, and permitting as space and energy capex start to collide.
Deep & Emerging TechPlug-in solar kits become legal in the UK tomorrow —but will you be buying one?
Household power is drifting from utility-scale to socket-scale — legalizing plug-in kits lowers the friction for renters and small properties to add generation without electricians or roof work. For energy and hardware operators, assume a fragmented, user-installed edge and design products and tariffs that treat every outlet as a potential micro-PV node.
Deep & Emerging TechPhotonic Publishes SHYPS QLDPC Code Results in Nature Communications
More efficient QLDPC error-correction — running larger programs on smaller quantum systems — is a quiet but real pull-forward on useful quantum timelines. If you’re tracking quantum risk or opportunity, start planning for software and algorithm work to matter earlier than hardware qubit counts alone would suggest.