The shift toward vertical integration in AI accelerates today as OpenAI unveils its first custom inference chip. We're also tracking Figma's aggressive move to bridge design and engineering natively on its canvas, plus the immediate turbulence hitting the fragile 60-day US-Iran peace roadmap.
OpenAI and Broadcom on Wednesday announced 'Jalapeño,' OpenAI's first custom-designed AI accelerator. Described as an 'Intelligence Processor,' the chip is purpose-built for large language model (LLM) inference and, according to the companies, delivers substantially better performance-per-watt than current state-of-the-art solutions. Developed in just nine months, Jalapeño is the first product in a planned multi-generation compute platform that OpenAI intends to deploy at gigawatt scale, marking a major step in the company's vertical integration from models down to silicon.
Why it matters
OpenAI's move into custom hardware is a significant strategic shift for the AI industry, signaling a future where leading labs design their own chips to optimize for specific model architectures and drive down inference costs. For a product builder, this trend toward specialized hardware could make advanced AI more accessible and efficient, potentially enabling new applications that are currently cost-prohibitive. The rapid development cycle, which OpenAI says was aided by its own models, also points to AI's growing role in designing its own underlying infrastructure.
Google announced on Thursday it has integrated native 'computer use' capabilities directly into its Gemini 3.5 Flash model. This built-in tool allows developers to create custom AI agents that can see the screen, reason about what's displayed, and take actions across web, mobile, and desktop environments. The goal is to enable autonomous operation within existing user interfaces, streamlining tasks like enterprise automation and software testing.
Why it matters
Baking 'computer use' directly into a model like Gemini 3.5 Flash is a significant step toward creating more capable and autonomous AI agents. For product builders, this unlocks the potential to build automation that interacts with legacy systems or third-party applications without needing dedicated APIs. It's a move to commoditize the ability for AI to operate software just as a human would, lowering the barrier to automating complex, multi-step digital workflows.
Fresh off the broad rollout of its AI Design Agent we tracked earlier this week, Figma used its Config 2026 conference on Wednesday to announce a suite of major updates aimed at bridging the gap between design and engineering. The flagship feature, 'Code Layers,' allows developers to integrate executable code from repositories directly onto the Figma canvas, creating interactive components that sync with production code. The company also launched 'Figma Motion' for native animation and timeline-based motion design, WebGPU-powered 'Shaders' for advanced visual effects, and expanded its AI 'Design Agent' with the ability for users to create custom, shareable plugins and skills using text prompts.
Why it matters
This is a significant strategic move by Figma to create a unified platform for the entire product development lifecycle, directly in your wheelhouse. For a design engineer, 'Code Layers' in particular could fundamentally change workflows by collapsing the traditional, often fraught, handoff between design mockups and live code. The ability to build interactive prototypes with real components and create custom AI tools directly within the design environment promises to accelerate iteration cycles and improve fidelity between what's designed and what's shipped.
Capping off a massive month for Cursor—which recently included a reported $60B acquisition by SpaceX and the purchase of open-source rival Continue—global IT services giant NTT DATA announced Thursday it is deploying the AI coding platform across its worldwide software engineering operations. The partnership aims to accelerate the modernization of legacy enterprise systems for its clients. NTT DATA will integrate Cursor's multi-model AI agents into its development environments, citing the platform's enterprise-grade governance features, including privacy controls and SSO, as key factors for the adoption.
Why it matters
This is a major validation for Cursor and the broader AI-native development tool space. The adoption by a massive services firm like NTT DATA signals that these platforms are moving beyond individual developer use to become core, strategic tools for enterprise-wide transformation. It highlights a clear industry shift towards AI-first engineering to tackle complex modernization projects at scale.
European fashion retailer Zalando is expanding its AI-powered size and fit tools across its markets to combat high return rates, which can reach 50% in online apparel. By leveraging machine learning, 3D body scanning technology, and customer feedback, the company aims to provide more accurate size recommendations. Zalando reports the initiative already prevented 8% of size-related returns in 2025.
Why it matters
Zalando's investment demonstrates a proactive strategy to tackle the massive operational and financial costs of returns in the fashion industry. Instead of penalizing customers with stricter policies, they are using technology to address a root cause of returns: poor fit. This approach is a prime example of using technology to improve both customer experience and reverse logistics efficiency, a key challenge in retail circularity.
Redo, a commerce technology company specializing in post-purchase experiences like returns and exchanges, has raised an $81 million Series B funding round, valuing the company at $1.25 billion. The investment, led by Smash Capital, will be used to advance its product, expand AI initiatives, and grow internationally. The funding follows Redo's recent acquisition of ReturnBear to streamline international returns.
Why it matters
This massive funding round for a returns-focused company underscores the intense investor interest in solving the costly and complex challenges of reverse logistics. For retailers, it signals that the technology for managing the post-purchase journey is becoming a critical part of the e-commerce stack, with a clear focus on using AI to improve customer retention and operational efficiency.
Spokane is the first city in Washington to utilize the state's 'Parking to People' incentive, a program designed to convert underused parking lots into affordable housing. The Kōz on West 4th development, a 210-unit project, is nearing completion and represents the first project to benefit, saving the developer $4.2 million through a sales tax exemption provided by the program.
Why it matters
This marks a concrete and successful application of a creative state policy to address Spokane's housing shortage. It provides a tangible model for urban revitalization that could be replicated elsewhere in the state, demonstrating a practical way to incentivize the construction of badly needed affordable housing on underutilized urban land.
Following a two-week closure for maintenance that we've been tracking, Spokane's Maple Street Bridge is reopening to traffic Thursday evening. Ahead of the vehicle reopening, Mayor Lisa Brown hosted a 'Bike the Bridge' event for pedestrians and cyclists to see the newly repaired infrastructure. The reopening comes just in time for the city's massive Hoopfest basketball tournament this weekend.
Why it matters
The timely reopening of this key arterial bridge is a significant logistical win for Spokane, averting major traffic congestion during Hoopfest, one of the city's largest annual events. It concludes a period of significant disruption for commuters and demonstrates the city's ability to manage critical infrastructure projects on a tight deadline.
The 60-day US-Iran peace roadmap we've been tracking is already showing significant strain, with both sides issuing explicit warnings over the Strait of Hormuz. Expanding on yesterday's public disagreements over terms, Iran's Revolutionary Guards (IRGC) warned Thursday they would take 'appropriate measures' against any vessels using unauthorized shipping routes. This followed statements from US officials, including Secretary of State Marco Rubio and President Trump, that any attempt by Iran to impose transit tolls—a major sticking point throughout negotiations—would be unacceptable. Meanwhile, conflicting reports persist about a potential Israeli withdrawal from Lebanon.
Why it matters
The public disagreement over the Strait of Hormuz, a critical chokepoint for global oil, highlights the deep mistrust and fragility of the recent peace framework. The deal's success hinges on resolving these core security and economic issues. The situation remains volatile, with the on-the-ground reality of vessel passage and military posturing being a key indicator to watch. Any miscalculation could quickly unravel diplomatic progress.
Google's A2UI v0.9, an 'Agent-to-UI' framework, introduces a new method for generating user interfaces. Instead of generating raw code, it produces structured JSON blueprints that reference a 'Trusted Catalog' of native components. This decoupled approach is designed to standardize UI creation across platforms by focusing on the developer's intent rather than direct code generation.
Why it matters
A2UI represents a significant conceptual shift for UI development, moving toward a more structured, declarative, and agent-friendly paradigm. For a design engineer, this framework could enforce design system consistency at a fundamental level, ensuring that AI-generated interfaces adhere to a pre-approved component library. It's a structured alternative to the 'vibe coding' approach of generating raw, unstructured front-end code.
A new open-source intelligence (OSINT) engine called Sightline has been released, designed to transform OpenStreetMap data into actionable geospatial intelligence. The tool allows users to query, visualize, and monitor over 200 types of physical infrastructure—from power grids to pipelines—using natural language. It aims to democratize access to infrastructure mapping for cybersecurity, supply chain analysis, and investigative journalism.
Why it matters
Sightline provides a powerful and accessible tool for mapping the physical world, an area of OSINT that often requires expensive commercial software. By leveraging open data, it allows analysts to quickly gain situational awareness of tangible assets, which is critical for everything from security reconnaissance and supply chain risk analysis to disaster response planning.
AI Industry Vertically Integrates with Custom Silicon Major AI labs are moving into hardware design to optimize performance and cost. OpenAI's partnership with Broadcom on the 'Jalapeño' inference chip and Qualcomm's new 'Dragonfly' CPU for agentic AI signal a trend toward custom, purpose-built silicon, reducing reliance on general-purpose GPUs and creating a tighter feedback loop between models and hardware.
Figma Blurs the Line Between Design and Code Figma's Config conference saw the launch of 'Code Layers,' allowing executable code to live directly on the design canvas. This move, along with native motion design and AI-powered plugins, aims to create a unified environment for product development, collapsing the traditional handoff between design and engineering.
Strait of Hormuz Becomes Key Point of Contention in US-Iran Deal Despite a framework peace agreement, the Strait of Hormuz remains a major friction point. Iran's IRGC is warning against unauthorized passage and seeking to control transit, while the US insists on free navigation. The disagreement highlights the deal's fragility and the potential for renewed disruption to global shipping.
AI Transforms Reverse Logistics and Returns Management Retailers are increasingly deploying AI to tackle the multi-billion dollar problem of product returns. Zalando is using AI fit tools to prevent size-related returns, while companies like Redo and SELLIT9 are attracting significant investment for AI-powered platforms that optimize the post-purchase and recommerce lifecycle.
Humanoid Robots Enter the Workforce The deployment of humanoid robots in logistics is accelerating. Figure AI's robots have now operated for 200 hours continuously, processing a quarter-million packages, while a partnership between Kawasaki and Dexterity is set to scale 'Physical AI' in warehouses. These milestones signal a move from pilot projects to scalable, autonomous labor in real-world environments.
What to Expect
2026-06-25—Maple Street Bridge in Spokane set to reopen to vehicle traffic.
2026-09-29—The AI Conference 2026 begins in San Francisco.
How We Built This Briefing
Every story, researched.
Every story verified across multiple sources before publication.
🔍
Scanned
Across multiple search engines and news databases
476
📖
Read in full
Every article opened, read, and evaluated
172
⭐
Published today
Ranked by importance and verified across sources
11
— The Anvil
🎙 Listen as a podcast
Subscribe in your favorite podcast app to get each new briefing delivered automatically as audio.
Apple Podcasts
Library tab → ••• menu → Follow a Show by URL → paste