Analysis

An AstraZeneca executive describes AI-driven, increasingly automated biologics design as oncology's longer-term engine. The source checks out at 0% of 11 claims. The hard evidence is elsewhere: Summit's SEC-verified filings show how costly the bispecific bet is before a Nov 14 FDA decision.

In the week to 14 October 2026, Cisco, Hesai, Zeta and TELUS Digital/Parloa each announced a different layer of the AI stack. Nvidia's SEC-checked filings show cost of revenue up more than 3.7x in two fiscal years. Most announcements are company press releases, a weaker source class, and we flag where.

SAP, Zeta and others announced agent platforms within days of each other in October 2026. The smaller deals show where the field is spending: Tiny bought the authorization company Oso Cloud for about $1.9 million in cash. EDB says policy must be enforced by the system, and a Norwest investor says AI shouldn't do finance calculations. Sources are cited, and two carry measured reliability scores.

AstraZeneca's Puja Sapra describes AI that designs molecules and robots that test them, just as oncology enters a dense run of regulatory and trial events. The AI claims come from one source whose checked claims did not hold up, so we separate what is verified from what is promised.

Micron's capital spending more than doubled in a year, to $15.9 billion in fiscal 2025, according to SEC-sourced figures. Nvidia, Bull and Kalray, and GlobalFoundries are all moving on memory supply and data movement. The sources behind these deals vary a lot in reliability, and we show that.

AstraZeneca's scientist argues that proprietary data, robots and AI will design the next biologics, but the article carrying her case scored 0% on our claim check. The only AI-adjacent oncology company with audited numbers here is Summit, whose FY2025 loss was $1.08 billion.

Nvidia is co-developing next-generation SK Hynix memory to secure HBM supply, while SEC filings show Micron's annual capex more than doubled in two years. Verified filings, plus some thinner, lower-reliability sources, show where the AI hardware squeeze is hitting.

Q4 2026 brings an FDA decision date for ivonescimab and the first detailed data for Merck and Moderna's personalised cancer vaccine. AstraZeneca is pitching AI-designed biologics to shorten discovery. We separate SEC-verified Merck figures from vendor-voice claims, and flag the AI source's weak fact-check record.

AstraZeneca describes AI that proposes, robots that test and humans who oversee, while mRNA vaccines and ADC-bispecific combinations head for key dates. Our dossier checks the AI claims against reliability data and Merck's filings. The AI claims rest on a source whose checked claims did not hold up.

SEC-filed figures show Micron's capital spending nearly doubled in fiscal 2025 and its cash reached about $25 billion by Q3 2026. Analyst enthusiasm for AI power and compute names runs well ahead of the evidence in places: D-Wave trades at 171x sales, and one source behind the NuScale call scores 22% on reliability.

AI demand is straining chip supply. SEC-filed figures show Micron's capital spending more than doubled in two years and its cash rose sharply. Deals and research are pushing toward new silicon. Several sources behind these stories score poorly on our reliability checks, and we flag where.

A survey reported by MIT Technology Review says every respondent plans to use agentic AI within two years, yet the money is flowing to governance, security and data plumbing. We flag that the survey's source scored 0% on our claim checks, and we ground the compute story in SEC-checked Nvidia figures.

A cluster of CEO handoffs runs from July to early October 2026. Almost none of it concerns AI. The one technology-adjacent filing, Kneat's, warns of vendors building rival validation tools. Source reliability scores of 30% and 57% suggest treating the glowing language with care.

Meta is building an enterprise platform under a departing MongoDB chief, and startup capital is flowing to narrow, vertical AI agents. Practitioners warn that policy must be enforced by the system, not the model, and that AI is poor at calculation. Every claim here traces to a named source, and we flag where those sources are weaker.

A survey says 100% of enterprises expect to use agentic AI within two years, yet only about 45% of company data is AI-accessible. We set that against Nvidia's checked SEC figures and CB Insights startup interviews. The survey source has a poor verification record, so we flag it.

Oncology's AI story rests on two tracks: personalized mRNA vaccines nearing detailed data, and AI-designed biologics at AstraZeneca. The AI claims are mostly aspirational and come from a source that scored 0% on our fidelity checks. The clinical and commercial signals are better sourced.

A survey says every enterprise respondent plans to use AI agents within two years, yet agents reach only about 45% of company data. Our check of the source found none of its 11 tested claims held up, so we treat the figures as unconfirmed. Governance-led partnerships and Nvidia's verified SEC figures show where spending is going.

SEC-filed figures show Micron's capital spending roughly doubled in fiscal 2025 and its cash pile reached $25.0B in fiscal Q3 2026. Around that hard data sit analyst claims of far weaker quality, and we flag which is which. The dossier holds no model or agent data, so this is a hardware-and-power story.

Banks, insurers and software vendors are committing to AI agents, and every respondent in one survey plans to use them within two years. The same survey says access to company data is the weak point. That source failed our reliability check, so we flag it.

Google now lets users remove visible watermarks from Gemini-made images, video and music — even as OpenAI loses another senior executive and AI infrastructure spending keeps compounding. Enterprise AI adoption data and a live model-compression benchmark show where the real, verifiable action is.