Where AI Is Actually Heading in 2026: Code, Cars, and Medical Scanners

The $60 Billion Question: Who Will Write the Code of the Future?

In April 2026, SpaceX stunned the tech world by signing a deal to either invest $10 billion in AI coding startup Cursor or acquire it outright for $60 billion in stock. Less than two months later, the acquisition is moving forward — and it will likely close before the end of the year. The deal is a seismic signal about where Silicon Valley thinks AI is headed.

Cursor has become the go-to AI-powered code editor for professional developers, powering everything from startup MVPs to enterprise codebases. Its appeal lies in its ability to understand context across entire repositories, suggest multi-file refactors, and generate production-quality scaffolding in seconds. For SpaceX, the purchase is not just about better internal tooling — it's about supercharging xAI, the artificial intelligence division SpaceX acquired from Elon Musk earlier this year. That division, which also merged with X (formerly Twitter), has been mired in controversy: a chatbot that briefly called itself "MechaHitler," the departure of all eleven original co-founders, and Musk's own admission that xAI "was not built right the first time around." A $60 billion infusion of Cursor's engineering talent and AI coding infrastructure is Musk's attempt to rebuild from the ground up — and it's the largest AI acquisition in history.

What makes this different from typical Big Tech acquisitions is the vertical integration. SpaceX isn't just buying a developer tool; it's buying a pipeline for training, testing, and shipping AI models at the scale required for Starship launches, Starlink constellation management, and eventually Mars mission software. The implications ripple across the entire software industry: if the company building your rockets also builds your AI coding platform, the barrier to entry for competing AI code assistants just got substantially higher.

Creative AI Comes of Age: From Sidebar Widgets to Brand Identity Engines

While Elon Musk is making billion-dollar bets on AI coding, Adobe is quietly turning its Firefly image-generation models into a full-fledged creative AI agent. In a major update announced this week, Adobe is integrating Firefly AI Assistant directly into Premiere Pro, Photoshop, Illustrator, InDesign, and Frame.io — not as a cloud overlay, but as a native sidebar panel inside each application.

The update represents a significant philosophical shift. Earlier versions of Firefly AI Assistant were useful but relied on users switching contexts to a separate web interface. Today's beta release brings the assistant into the tools professionals already use daily. More importantly, the assistant now maintains cross-session context: a filmmaker working on a campaign in Premiere can ask the assistant to generate a storyboard, and later in Photoshop those same visual elements will be recognized and extended. A new "Elements" feature lets users save characters, locations, and objects for reuse across projects — effectively giving the AI a persistent memory of a brand's visual identity.

New creative skills include the ability to generate a rough video cut from a handful of clips, produce video from storyboards, and develop a complete brand identity that carries styling, logo concepts, and color palettes forward into every generated asset. In a demo, an Adobe employee used the assistant inside Illustrator to randomize the placement and color of hundreds of circles — a tedious task that previously required manual work or complex scripting. The assistant completed it in seconds.

Adobe cautions that Firefly AI Assistant is not a "computer-use agent" that takes over your cursor; the company says its focus is on delegating repetitive work so creatives can spend more time on creative decisions. Yet the trajectory is clear: the line between "AI tool" and "AI collaborator" is blurring fast, and the professional creative suite is where that blur is most visible.

The Car Dashboard Wars: Why Automakers Are Ditching Android Auto

For roughly a decade, the relationship between smartphone and car was simple: you plug your phone into the car, Android Auto or Apple CarPlay takes over the infotainment screen, and everyone is reasonably happy. That handshake agreement is unraveling — and it's one of the most underappreciated tech stories of 2026.

The turning point came when General Motors announced it was dropping Android Auto and Apple CarPlay from its electric vehicle lineup, replacing them with a proprietary conversational infotainment system built on Google's Android Automotive OS (AAOS) but powered by Gemini AI. GM made clear the decision wasn't about cutting ties with Google — it was about data sovereignty. With phone projection, all the GPS, mapping, and driving behavior data flows to your phone. GM wants that data inside its own systems so it can improve EV range estimation, optimize charging routes, and integrate with its Super Cruise driver-assistance system.

The broader context matters. Google doesn't charge automakers for Android Auto integration, but it does harvest enormous amounts of behavioral data — where you drive, how you drive, where you stop, what you listen to, and increasingly, what you ask your AI assistant. In 2023, GM paid a $12.75 million fine to California for misusing customer driving data, which may have sharpened the company's appetite for building its own walled garden. Other manufacturers that never adopted phone projection — notably Rivian and Tesla — have been watching, and several legacy brands are now reconsidering their own in-house OS strategies.

Meanwhile, Google is fighting back. At I/O 2026, the company unveiled a major Android Auto redesign with Material 3 Expressive UI, customizable widgets, immersive 3D navigation with traffic light and lane-level detail, and deeper Gemini integration — including Magic Cue, a contextual assistant feature that surfaces addresses, appointments, and other information proactively while you drive. For cars with Google built-in (Android Automotive), Gemini can now explain dashboard warning lights, tell you whether that couch you're buying will fit in your trunk, and use onboard cameras to improve lane-level navigation accuracy.

The titan vs. upstart dynamic is familiar, but the stakes are different this time. Car infotainment is becoming the next computing frontier, and both sides know it.

Unreal Engine Goes Gen AI: Game Development's Crossroads

At Unreal Fest, Epic Games gave an extensive look at how it's embedding generative AI into Unreal Engine 6 — and revealing the growing tension within the game industry about AI's role. Despite a recent GDC survey finding that 52 percent of game developers believe generative AI is bad for the industry (up from 30 percent the previous year), Epic is moving full speed ahead.

The centerpiece is an experimental Model Context Protocol (MCP) plugin shipping in Unreal Engine 5.8 today, with plans to make MCP a foundational part of UE6. The plugin allows developers to connect AI models — specifically Claude and Gemini — directly to the engine. In a live demonstration, Epic showed Claude Code pulling objects from an asset library and placing them inside a virtual living room, building and adjusting city layouts as parks and buildings were added, and modifying lighting and atmospheric conditions to match real-world reference images. The AI can access core UE systems including blueprints, assets, levels, materials, and meshes.

Epic's CEO Tim Sweeney had previously suggested that a "made with AI" tag would soon be irrelevant because "AI will be involved in nearly all future production." The MCP plugin is the infrastructure for that future. Epic describes the AI models as "creativity and productivity multipliers" that reduce repetitive manual work, letting teams iterate faster and focus on creative decisions. Yet the developer response remains deeply ambivalent. More than a third of surveyed developers report using gen AI for research and brainstorming, but the proportion who see it as harmful to the industry has nearly doubled in a single year. Epic's bet is that developers who embrace the tool will outpace those who don't — and that the tools themselves will evolve to win over skeptics.

The Wildcard: Midjourney's Full-Body Ultrasonic Medical Scanner

If you had "AI image generator builds $60,000 medical imaging machine" on your 2026 bingo card, congratulations — you can mark it off. Midjourney, the company famous for its text-to-image tool, has announced its first hardware product: a full-body ultrasonic scanner that produces a 3D map of a human body down to a fraction of a millimeter in under 60 seconds.

The announcement was so unexpected that early readers assumed it was an April Fools' joke. It isn't. Midjourney is founding Midjourney Medical to develop the scanner in partnership with Butterfly Network, the handheld ultrasound device maker with which it signed an exclusive licensing agreement in late 2025 for ultrasound-on-chip technology. The project is led by Ahmad Abbas, Midjourney's head of consumer hardware — a former Apple Vision Pro engineer who joined the company in 2023.

The scanner works by submerging the user in water at a controlled rate while they stand on a platform. The body passes through a ring containing half a million tiny ultrasonic transducers — each the size of a grain of sand — that emit waves and record echoes bouncing back. Midjourney likens the experience to being surrounded by half a million tiny dolphins, using echolocation to produce a full 3D body map at nearly 100 times the speed of a conventional MRI (which typically takes 60 to 90 minutes).

The roadmap is aggressive. Over the next 12 months, Midjourney will refine algorithms, conduct research trials, and build a second-generation prototype. It plans to open its first "Scanner Spa" in San Francisco in 2027 — a hybrid medical facility and wellness center where customers can walk in and get scanned. FDA diagnostic approval is the critical next step. By 2028, Midjourney aims for a third-generation machine using custom silicon for dramatically improved image quality, with expansion to more cities. The company's long-term target is 50,000 scanners worldwide by 2031.

The ambition is staggering. Midjourney claims that widespread early imaging could potentially avoid 30 percent of all deaths and cut 50 percent of healthcare costs. Even allowing for Silicon Valley optimism, a sub-minute full-body scan at near-MRI resolution — if it works as described — would fundamentally change preventive medicine, oncology screening, and emergency diagnostics. It also represents the most dramatic pivot in tech history: from whimsical image generation to medical hardware at scale.

AI Regulation in the Classroom: Norway's Bold Stand

While corporate America races to build bigger AI products, some governments are building guardrails. Norway is imposing one of the strictest AI-in-education policies in the world, effective at the start of the 2026-27 school year in late August. The policy restricts generative AI use in schools by age group: children aged 6 to 13 (grades 1 through 7) will be effectively banned from using AI tools; students aged 14 to 16 may use AI only under direct teacher supervision; and those aged 17 to 19 are encouraged to use AI "appropriately" in preparation for higher education and the workforce.

Prime Minister Jonas Gahr Støre argued that AI tools allow children to skip essential steps in learning to read, write, and do mathematics — the foundational skills the government believes schools should prioritize. The policy follows Norway's 2024 ban on smartphones in classrooms, which the government credits with reduced bullying, better grades, and fewer mental health visits, particularly among girls. A planned social media ban for all children under 16 is also being drafted for parliamentary introduction by the end of 2026.

In the US, Congress is separately advancing the GUARD Act (Guidelines for User Age-verification and Responsible Dialogue), which would require AI companies to implement age-verification systems and restrict AI chatbots for minors. The bill passed the Senate Judiciary Committee but remains unvoted. Its language recently narrowed from covering nearly all AI chatbots to focusing on "AI companions" — a change that critics argue could exempt major products like ChatGPT, Gemini, and Copilot if their chatbot functions are deemed incidental to their primary purpose.

The global regulatory landscape is fragmenting, but the direction is consistent: AI in education and for minors is becoming a policy battleground, and companies that build consumer AI products will increasingly face jurisdictional compliance requirements that could slow international rollout.

The Thread That Connects Them

These stories may seem unrelated — a $60 billion space-company acquisition, a design tool sidebar, a car dashboard war, a game engine plugin, a medical scanner, and a Norwegian school policy — but they share a common theme. In 2026, AI is no longer a feature; it is the substrate. Every major software platform, hardware device, industry vertical, and regulatory framework is being rebuilt around AI as the default computational primitive.

The coding tools will write the software. The car dashboards will drive the routes. The medical scanners will catch the diseases. The game engines will build the worlds. The classroom policies will determine who learns to think with AI and who learns to think despite it. The companies that understand this — not just as a trend to ride but as a fundamental restructuring of how value is created — are the ones making the biggest bets. The rest are still catching up.

---

Sources: Engadget, The Verge, Reuters. Published June 21, 2026.