the Now & the Next

A bi-weekly speculative fiction suggesting the shape of things to come.
(sourced from trustworthy trade pubs, think tanks + frontier science news)

12,093 Signals Tracked
7 Collisions Detected
6 Sectors Spanned
308 Publications
22 Day Signal Window

Three weeks of signal. Twelve thousand data points. What the longer view reveals: disruption is moving in parallel across sectors that don't talk to each other — beauty counters, combat aircraft, solar physics labs, ad server logs, university registrar offices, firewall dashboards, and the Strait of Hormuz. These collisions aren't predictions. They're already in progress.

01
The Beauty Intelligence Race

L'Oréal, Sephora, and Ulta aren't waiting for AI to mature. They're deploying it against each other.

+

L'Oréal deepened its OpenAI partnership to embed autonomous agents across research, marketing, and consumer experience. Ulta Beauty launched an AI chatbot powered by Google Gemini, integrated across its site and app. Sephora embedded its shopping experience inside ChatGPT and deployed a proprietary AI Beauty Chat. Fenty Beauty began building a strategy to rank inside agentic AI platforms — because traffic from ChatGPT is already measurable. Jo Malone London launched Scent Scanner on Pinterest: an AI fragrance recommendation tool that builds a personalized wardrobe. Meanwhile, Haut.AI debuted SkinGPT at VivaTech — generative AI that simulates how your skin will look in ten years. The beauty industry has become the most aggressive enterprise AI adopter in consumer goods, moving faster and with more household name visibility than any other sector. When five of the world's most recognized beauty brands deploy AI in a single fortnight, it stops being experimentation and becomes competitive infrastructure.

⚡ The Now

Beauty brands are using AI to collapse the distance between desire and purchase. Every touchpoint — discovery, personalization, recommendation, checkout — is being intermediated by AI agents. L'Oréal is automating research and creative effectiveness measurement. Ulta's chatbot is learning continuously from in-session behavior. Sephora's ChatGPT integration serves product recommendations through the same interface consumers use to ask about everything else. Fenty's team is optimizing not for Google rank, but for how AI systems describe the brand. The shift is from 'AI for efficiency' to 'AI as the consumer relationship itself.'

→ What's Next

The beauty counter is becoming an AI conversation. Within 18 months, the brands that have instrumented their product catalogs for AI retrieval will have a structural advantage over those still optimizing for human search. SkinGPT points further: the next form of beauty marketing isn't advertising — it's predictive visualization of your future self, tied to a product recommendation. The brands that win the next decade of beauty won't have the best products. They'll have the best AI personas.

Consumer Goods Technology
L'Oréal has partnered with OpenAI to embed agentic commerce capabilities, enhance consumer journeys, and automate research and marketing functions across its global operations.
Glossy
Ulta Beauty launched an AI chatbot powered by Google Gemini, integrated into its on-site and in-app experience, with continuous enhancement based on consumer interactions.
Glossy
Sephora integrated its shopping app within ChatGPT, introduced proprietary AI Beauty Chat, and deployed AI-enabled in-store tools — a multi-platform AI strategy.
Chief Marketer
Fenty Beauty is building a strategy to improve its visibility in agentic AI platforms, driven by measurable consumer traffic growth from ChatGPT in 2026.
Cosmetics Business
Haut.AI launched SkinGPT, generative AI that produces photorealistic simulations of skin evolution over time — allowing brands to show consumers their future skin.
02
The Autonomous Sky

The Air Force just chose its first drone wingmen. Human pilots are no longer the default architecture of air combat.

+

The US Air Force selected General Atomics and Anduril to build its first Collaborative Combat Aircraft — AI-piloted drones designed to fly alongside and in front of human-crewed fighters. Honeywell Aerospace immediately announced plans to enter the global CCA market. Poland is evaluating the X-BAT autonomous vertical-takeoff fighter for its air force. Ukraine's defense AI chief predicted a 'new paradigm' of warfare driven by autonomous systems. Across three weeks, the pattern is unmistakable: the air combat architecture that has defined military doctrine since World War I — pilot in cockpit — is being redesigned around AI systems that don't breathe, don't tire, and don't fear. The CCA selection isn't an experiment. It's a production contract. The drone wingman era has arrived.

⚡ The Now

AI-piloted aircraft are moving from concept to contract. The Air Force CCA program represents the first time autonomous combat aircraft will be built at production scale, not as prototypes. Anduril's selection signals the new defense prime — software-first companies displacing legacy primes for next-generation platforms. Poland's X-BAT evaluation shows this is a NATO-wide architectural shift, not just a US program. Ukraine's battlefield AI is producing doctrine in real time: swarm tactics, autonomous reconnaissance, AI-coordinated artillery are no longer theoretical.

→ What's Next

The air dominance equation is being rewritten. Within a decade, the ratio of AI-to-human aircraft in advanced air forces will invert. The industrial implication: defense contractors that can't build software-defined, autonomy-first platforms will lose market share to Anduril-class companies. The strategic implication: countries that field autonomous air systems at scale will dictate the terms of 21st-century deterrence. The ethical implication — who authorizes a lethal autonomous action — has no agreed answer yet.

Air & Space Forces Magazine
The US Air Force awarded production contracts to General Atomics and Anduril to build Collaborative Combat Aircraft — autonomous drone wingmen designed to operate alongside crewed fighters.
Breaking Defense
The CCA selection marks the transition from prototype to production-scale autonomous combat aircraft, with Anduril and General Atomics splitting the initial contracts.
C4ISRNET
Poland is evaluating acquisition of up to 32 autonomous vertical-takeoff fighters as part of a defense modernization effort, reflecting broader NATO interest in autonomous air systems.
C4ISRNET
Ukraine's Defense Ministry is implementing AI-driven systems across recruitment, operations, and battlefield coordination, with its defense AI chief calling the result a 'new paradigm' of warfare.
03
The Solar Reinvention

The next decade of solar gains won't come from more panels. They'll come from stranger ones.

+

LONGi unveiled what it calls a 'sunlight generator' at SNEC 2026 — an integrated solar-plus-storage system designed to make solar dispatchable. Researchers at Nanjing University built a floating panel that harvests energy from both sunlight and rain. Fraunhofer ISE introduced ShadeCut: photovoltaic technology camouflaged to look like roof tiles and brick walls, designed for historic buildings that could never install traditional panels. Scientists at Kyushu University achieved a 130% quantum yield in solar energy conversion — extracting more energy from photons than was theoretically possible under single-junction physics. Tongwei launched modules with annual degradation rates so low they will outlast the warranties that defined the previous generation. Across three weeks, solar technology crossed a threshold: it stopped being an engineering problem of 'how many panels' and became a materials science problem of 'what can a panel do.' The answers emerging from labs and factory floors are genuinely strange.

⚡ The Now

Solar is entering a materials science era that the installation industry hasn't priced in yet. LONGi's sunlight generator reframes solar from an intermittent power source to a firm power product — the storage integration changes the economics of every grid connection. The Fraunhofer camouflage panels open markets that have been legally closed: historic preservation zones in Europe, urban cores in Asia, high-density neighborhoods everywhere. The rain-harvesting panel points toward all-weather generation. None of these are decades away. SNEC 2026 showed them at production-ready stages.

→ What's Next

The solar panel of 2030 will be unrecognizable to the installer of 2024. Quantum yield breakthroughs, multi-modal energy harvesting, and integrated storage will converge into panels that generate more energy than physics once permitted, in places they were never allowed, without the intermittency problem that made them second-class power citizens. The policy implication: grid planning built on the assumption of solar as variable generation will need to be rewritten. The investment implication: materials science companies — not traditional solar manufacturers — may capture the next wave of value.

Energies Media
LONGi unveiled its integrated solar-plus-storage LONGi ONE strategy at SNEC 2026, positioning the system as a dispatchable 'sunlight generator' rather than an intermittent power source.
Energies Media
Researchers at Nanjing University developed the W-DEG, a floating panel that harvests energy from both sunlight and rainfall, enabling all-weather generation on water surfaces.
Energies Media
Fraunhofer ISE introduced ShadeCut, a photovoltaic technology that camouflages solar panels as building materials — opening historic preservation zones and urban cores to solar generation.
Energies Media
Kyushu University researchers demonstrated a novel molybdenum-based spin-flip process achieving 130% quantum yield — extracting more energy from photons than single-junction physics previously allowed.
04
The Adtech Rewrite

AI agents are replacing human media buyers. The advertising industry's operating system is being rebuilt layer by layer.

+

The Trade Desk launched Koa Agents — AI systems that automate media buying end-to-end based on campaign goal inputs. Google Ad Manager launched an AI chatbot for publisher ad operations teams to troubleshoot campaign delivery without human specialists. PMG, an independent agency, adopted the new MCP integration to enhance YouTube targeting — the same protocol architecture that connects AI models to enterprise tools is now connecting AI to ad servers. Hearst Magazines launched Aura IQ, an AI-driven platform that automates audience-building and campaign planning. ITV rolled out GenAI Ads Manager for AI-generated ad creative at scale. A WPP Media report projected generative AI search advertising revenues at £75 billion by 2030 — growing from 2% of intelligence category revenues to 39% in five years. The advertising supply chain was already opaque. Now it is being replaced by agents that make decisions faster than humans can audit them.

⚡ The Now

The human media buyer is being intermediated out of the buying loop. Koa Agents and Google's chatbot represent different ends of the same shift: on the buy side, AI sets strategy and executes; on the sell side, AI troubleshoots and optimizes. MCP in advertising means any AI model can now call ad server APIs directly — the agency as middleware is optional. The £75 billion projection for AI search ads isn't speculative; generative AI is already serving ads inside ChatGPT and Gemini, and the measurement infrastructure to track it is being built now.

→ What's Next

By 2028, the most consequential media buying decisions will be made by agents, not planners. The agencies that survive will be the ones that own the AI configuration layer — the goal-setting, constraint-setting, measurement frameworks that agents execute against. The publishers that win will be those who have instrumented their inventory for AI retrieval and agent-compatible APIs. The brands that lose will be those that still think the relationship is between their CMO and a human account team.

AdExchanger
The Trade Desk launched Koa Agents, enabling marketers to specify campaign goals while AI automates media buying workflows — a rules-based, goal-directed agentic buying system.
AdExchanger
Google Ad Manager launched Ask Ad Manager, an AI chatbot for publisher ad ops teams to troubleshoot campaign delivery issues without requiring specialist intervention.
AdExchanger
PMG adopted MCP integration to connect AI models directly to YouTube's ad infrastructure — enabling AI-driven targeting improvements without human intermediation.
Marketing Week
WPP Media projects generative AI search advertising will grow from 2% of revenues in 2026 to 39% by 2031, reaching £75 billion — a structural shift in where ad budgets flow.
Adweek
Hearst Magazines launched Aura IQ, automating audience-building, campaign planning, and optimization across its portfolio of magazine brands.
05
The Credential Scramble

Universities are redesigning degrees in real time. The labor market is already two years ahead.

+

The University of Miami overhauled its entire business school curriculum to embed AI across every course and program. Virginia Tech launched a new interdisciplinary AI minor pulling students from computer science, engineering, art, business, and social science simultaneously. Connecticut colleges introduced new AI bachelor's degrees and certificate programs to meet employer demand. Rockwell Automation reported that manufacturers are using AI to capture and transmit expert knowledge — not to replace workers, but because experienced workers are retiring faster than they can be replaced. Research from ICIMS showed demand for AI-skilled hires rising year-over-year despite tech sector layoffs. Lawmakers are openly debating whether higher education can move fast enough. The gap between what AI can do and what credentials certify is widening every quarter. Universities are running at course-revision speed. The labor market is running at model-release speed.

⚡ The Now

The credential is decoupling from the competency. A student who graduated in 2024 with a data science degree is already working with AI tools that didn't exist when the curriculum was designed. Virginia Tech's interdisciplinary AI minor is a structural acknowledgment that AI fluency can't be contained in a single department. Miami's full curriculum overhaul is rarer — and signals how far behind a traditional MBA had already fallen. The ICIMS demand data is the market verdict: employers have stopped waiting for credentials to catch up.

→ What's Next

The four-year degree will not survive this cycle unchanged. Within five years, the institutions that redesign around demonstrated AI competency — not just AI coursework — will attract the students and employer partnerships that define institutional relevance. The institutions that add an 'AI module' to existing curricula will produce graduates who look current on paper and obsolete in practice. The parallel disruption: bootcamps, micro-credentials, and AI-native learning platforms are filling the gap faster than universities can accredit new programs.

Higher Ed Dive
The University of Miami Herbert Business School overhauled its curriculum to embed AI extensively across all courses and programs, including new AI-focused degree tracks.
Times Higher Education
Virginia Tech developed a new interdisciplinary AI minor integrating students from computer science, engineering, art, business, and social sciences — treating AI fluency as cross-disciplinary infrastructure.
HR Dive
ICIMS research shows year-over-year growth in employer demand for AI-skilled workers, even as headline tech layoffs continue — the AI-capable labor shortage is independent of the broader tech cycle.
Route Fifty
Policymakers and higher education leaders are debating whether universities can adapt curricula fast enough to match AI's pace of capability development.
06
The Fortinet Moment

75,000 firewalls exposed. A CVE chain in an AI gateway handed attackers admin. The infrastructure is failing.

+

Researchers uncovered FortiBleed — a campaign targeting 75,000 Fortinet FortiGate firewalls worldwide, enabling threat actors to extract configuration data and credentials from enterprise networks at scale. Varonis disclosed SearchLeak (CVE-2026-42824), a vulnerability in Microsoft 365 Copilot Enterprise Search that allowed data theft through a crafted Teams message. A three-CVE chain in LiteLLM — the AI gateway proxy deployed by thousands of enterprises — escalated privileges to full admin. Novo Nordisk suffered a prolonged breach in which clinical trial data and intellectual property were stolen. The US federal government's cyber grant program for states and localities is facing an uncertain future. Iran's cyber threat to US infrastructure persists despite the preliminary Hormuz peace deal. The pattern across these incidents is consistent: the attacks aren't hitting the AI models. They're hitting the middleware and infrastructure that connects AI to enterprise data — the layer nobody audited.

⚡ The Now

The attack surface has moved from endpoints to AI integration layers. LiteLLM is used by thousands of enterprises to route prompts between AI models and corporate tools. A vulnerability in that proxy doesn't compromise one company — it compromises the abstraction layer beneath many. SearchLeak shows the same dynamic: Copilot's connection to email and document systems creates an exfiltration path that bypasses traditional data loss prevention. FortiBleed at 75,000 firewalls suggests the underlying network security layer was never updated to assume AI-connected systems. Security teams are auditing models. Attackers are auditing the connective tissue.

→ What's Next

The next major enterprise breach will be blamed on an AI tool, but the root cause will be the integration layer beneath it. MCP, A2A, and similar protocols are rapidly expanding the attack surface by connecting AI agents to corporate systems without the security review that traditional software integrations required. The companies that will survive this cycle are those treating every AI integration point as a new trust boundary requiring explicit security design. The regulatory response — whether EU AI Act enforcement, SEC disclosure requirements, or sector-specific mandates — will arrive after the first major incident that can be cleanly attributed to an AI-adjacent middleware failure.

Network World
The FortiBleed campaign targeted 75,000 Fortinet FortiGate firewalls globally, extracting configuration files and credentials from enterprise networks at unprecedented scale.
VentureBeat
SearchLeak (CVE-2026-42824) in Microsoft 365 Copilot enabled data exfiltration via a Teams message, while a three-CVE chain in LiteLLM allowed full admin escalation — both attacks targeting AI integration layers, not models.
Fierce Pharma
FulcrumSec claimed a two-month breach of Novo Nordisk, stealing clinical trial data and intellectual property — one of the most sensitive categories of enterprise data.
Governing
The $1 billion federal cybersecurity grant program for state and local governments faces an uncertain future, leaving critical infrastructure security underfunded at the moment it is most needed.
Nextgov
US intelligence officials assess that Iran's cyber operations targeting American networks continue unabated despite the Hormuz diplomatic agreement.
07
The Maritime Risk Stack

Hormuz tolls, Russian warning shots in the English Channel, Ukrainian grain under fire. The world's shipping arteries are under simultaneous pressure.

+

Iran announced it will charge commercial ships transiting the Strait of Hormuz after the 60-day toll-free grace period expires — converting the world's most critical oil chokepoint from a military flashpoint into a permanent toll booth. Days earlier, Iran's IRGC launched drones at commercial ships in Hormuz hours after signing a peace agreement. A Russian Navy frigate fired warning shots at a British yacht in the English Channel — the first such incident in decades. Russia intensified missile and drone attacks on Ukrainian seaports, with projections showing a potential 33% reduction in Ukraine's monthly grain exports. Nearly 18,000 Indian mariners remain stranded in the Persian Gulf. The Baltic and International Maritime Council reported that despite the US-Iran deal, shipping insurers are not reducing war risk premiums. Three of the world's major maritime corridors — the Gulf, the English Channel, the Black Sea — are under active threat simultaneously. This is not coincidence. It is the shape of a world in which great power competition has no agreed rules of engagement at sea.

⚡ The Now

Global shipping insurance is pricing in a sustained period of multi-theater maritime risk. The BIMCO assessment — that the deal doesn't quell the risk — is the market's verdict: insurers who price physical risk are not reassured by diplomatic language. The Hormuz toll announcement converts a military risk into a structural cost: every barrel of oil transiting the strait will carry a new rent to Iran indefinitely. The English Channel incident and the Ukrainian port attacks are not connected by alliance, but they share a logic: maritime rules of engagement that held for 70 years are being tested in multiple theaters at once.

→ What's Next

The era of frictionless global shipping is ending. If Hormuz tolls become permanent, they will be built into the cost of oil, LNG, and the manufactured goods that transit through Asian supply chains. Routing alternatives — the Cape of Good Hope, Arctic passages, land corridors — will absorb some traffic but cannot substitute at scale. The companies that will be affected first are those with just-in-time supply chains built on the assumption that maritime routes are free and safe. The nations that will feel it most are those dependent on food and energy imports through a single chokepoint.

Marine Insight
Iran announced it will implement tolls on commercial ships transiting the Strait of Hormuz after a 60-day grace period — converting the world's most critical oil chokepoint into a permanent revenue stream.
Marine Insight
Iran's IRGC launched drones targeting commercial ships in Hormuz hours after signing a peace agreement with the US — demonstrating the gap between diplomatic text and ground reality.
Marine Insight
A Russian frigate fired warning shots near a British yacht in the English Channel on June 16, 2026 — the first such incident in the Channel in decades.
The Dairy Site
Russian attacks on Ukrainian seaports and energy infrastructure are projected to cut Ukraine's monthly grain exports by up to 33%, threatening global food supply chains.
Business Insurance
BIMCO reported that despite the US-Iran diplomatic deal, shipping insurers are maintaining elevated war risk premiums — the market's assessment that the deal does not resolve the underlying risk.

Frontier Science Feeding the Machine

Ten signals from research labs that don't make enterprise headlines — but are quietly building the substrate of what comes next.

Solar Physics
Researchers achieved a significant efficiency boost in flexible perovskite solar cells using uniform submicron pyramid surface structures — opening new form factors for solar beyond rigid panels.
Compute Architecture
IEEE Spectrum examines sound-driven synaptic computing — neuromorphic architectures that could dramatically reduce the energy cost of AI inference without sacrificing performance.
AI Infrastructure
KAIST researchers developed a liquid cooling system ten times more efficient than current AI chip cooling — directly addressing the thermal bottleneck limiting data center density.
Data Storage
The TextaDNA project advances DNA-based data storage using synthetic polymer fibers — a potential path to archival storage with density and longevity that silicon cannot approach.
Environmental Science
Researchers have mapped microplastic distribution in living human tissue at the cellular level for the first time — providing the first direct evidence of cellular-level contamination.
Vaccine Science
An mRNA-based influenza vaccine demonstrated superior immune response breadth and duration compared to traditional flu vaccines — extending the mRNA platform beyond COVID applications.
Drug Discovery
AI models trained on condensate morphology can now predict gene regulatory activity — a novel approach to drug target identification that bypasses traditional genomic sequencing.
Gene Therapy
Intellia Therapeutics reported phase 3 results for a one-time CRISPR-based treatment for hereditary angioedema — a potential model for permanent genetic correction of inherited diseases.
Climate Science
Researchers developed an early warning system for lethal river heatwaves affecting freshwater fish populations — the first predictive tool for aquatic thermal stress events.
AI Models
Hugging Face researchers benchmark alternative fine-tuning approaches against LoRA, the dominant technique for adapting large language models — raising questions about whether the default is still optimal.