Google rents SpaceX AI GPUs & Apple’s Core AI and Siri - Tech News (Jun 9, 2026)

Google is reportedly about to rent an enormous slab of AI compute from SpaceX—at a price tag so large it tells you the real AI race isn’t just models anymore, it’s raw power and hardware. Welcome to The Automated Daily, tech news edition. The podcast created by generative AI. I’m TrendTeller, and today is June 9th, 2026. Let’s get into what matters in tech right now—and why it’s interesting.

Google rents SpaceX AI GPUs

Let’s start with that infrastructure stunner. Reports say Google has signed a deal to pay SpaceX around nine hundred and twenty million dollars per month for AI compute capacity, tied to a huge fleet of Nvidia GPUs plus supporting hardware. If this holds, it’s a loud signal that even the biggest cloud players can hit a wall when demand outruns how fast you can build data centers, secure grid power, and rack machines. It also hints at a new business shape: companies that assemble serious compute can lease it like industrial real estate—and suddenly become “AI landlords.”

Apple’s Core AI and Siri

Staying in the AI power lane, Nvidia and SK hynix announced a multiyear partnership aimed at next-generation memory for scaling what Nvidia calls AI factories. This isn’t a flashy consumer story, but it’s a practical one: memory speed and supply increasingly decide how fast AI systems can run and how many you can deploy. The collaboration also leans into using AI and simulation to speed chip design and even model factory operations, which speaks to a broader trend—automation creeping deeper into how hardware itself gets built.

OpenAI moves toward IPO

Now to Apple, which had a busy stretch of AI news. First, Apple published beta documentation for Core AI, a new framework meant to let developers run AI models directly inside apps across Apple platforms, tuned for Apple silicon. The takeaway isn’t the API details—it’s the direction: Apple is trying to make on-device AI feel like a standard app capability, not a special project. That can enable faster features, lower reliance on servers, and potentially better privacy—because more work can stay local on your device.

Google Search thrives with AI

Apple also detailed a major redesign of Apple Intelligence, including a new foundation-model architecture that it says was co-developed with Google using technology behind the Gemini family. Apple’s pitch is that it can mix on-device processing with Private Cloud Compute for heavier tasks, while keeping tight privacy controls. Alongside that, Apple previewed a more conversational Siri slated for later this year, plus AI features that plug into everyday apps like Camera and Safari. The interesting subtext here is Apple’s balancing act: it wants to catch up in capability, but it’s still selling trust and device integration as the reason to choose its AI over everyone else’s.

Nvidia and SK hynix memory pact

Over on the business side of AI, OpenAI has reportedly filed confidential IPO paperwork with the U.S. SEC, a first formal step toward going public—though the company says it hasn’t decided when, or even if, it will pull the trigger. An OpenAI IPO would be a major test of investor appetite for the most visible name in consumer AI at a time when training and running models demands staggering amounts of capital. It also fits the current vibe: AI isn’t just competing on clever software anymore; it’s competing on who can finance and secure the infrastructure to scale.

Breach disclosures getting slower

Meanwhile, fears that ChatGPT would gut Google Search haven’t played out the way many predicted—at least not yet. Alphabet says AI features are pushing overall query volume to new highs and improving ad performance for marketers. But there’s a tradeoff: AI summaries can reduce click-through to publishers, and even with opt-out tools, many sites feel they can’t risk losing visibility. The headline is that Google’s distribution muscle still matters more than any single chatbot—especially when AI is baked directly into the place most people already go to ask questions.

UK pushes device child protections

Shifting to cyber and safety, Have I Been Pwned just logged its thousandth breach, and Troy Hunt used the milestone to argue that breach disclosure delays are getting worse—despite years of privacy regulation. His point is simple: companies often wait weeks to notify people, even when basic early warnings could help victims protect accounts. A related supply-chain security report from Black Kite adds another uncomfortable detail: attackers are increasingly exploiting vulnerabilities before those flaws are even publicly disclosed, and access can be passed to other criminals almost instantly. Put together, it’s a reminder that speed is now the currency in security—speed to detect, speed to warn, speed to respond.

China approves brain-computer implant

In the UK, Prime Minister Keir Starmer has given major tech firms—named were Apple and Google—until September to add system-level tools that can block children from viewing or sharing sexually explicit images on phones and tablets. The government is framing it as a direct response to rising child exploitation referrals, while civil liberties groups warn that device-level blocking and age checks could chip away at privacy and anonymity. This is one of those moments where governments are trying to push safety controls down into the operating system itself, and the outcome will shape how far platforms are expected to police content at the device layer.

New warnings on nuclear risks

From China, there’s notable progress in brain-computer interfaces: reports say China approved a brain implant called NEO for commercial sale after clinical trials, aimed at helping people with paralysis and spinal cord injuries. It’s being compared to Neuralink, which is still in limited human testing. Beyond the competitive angle, the bigger story is what comes next: if BCIs move toward wider deployment, the privacy and security stakes get extreme. Neural data isn’t just another biometric—it’s intensely personal signals, and it will force new rules about consent, storage, and abuse prevention.

StormWall concept for solar storms

Two more quick science-and-risk notes to close. First, SIPRI’s Yearbook 2026 warns that nuclear-armed states are increasingly treating nuclear weapons as tools of national power, with modernization accelerating and transparency declining after the expiry of key arms-control limits. Even for a tech audience, this matters: escalation risk is also a systems risk—satellites, communications, and critical infrastructure all sit downstream of geopolitical stability. And finally, researchers proposed a concept called StormWall: a space-based system that would release material that turns into plasma to reduce the impact of extreme solar storms. It’s still a proposal, but it’s an example of how dependent we’ve become on space and electricity—and how seriously people are starting to treat once-in-a-generation geomagnetic events as something worth engineering defenses against.

That’s the tech landscape for June 9th, 2026: AI demand pushing infrastructure into blockbuster contracts, platforms hardening their on-device AI strategies, regulators pressuring operating systems to do more, and security teams racing a faster threat timeline. If you want to support the show, share this episode with someone who follows AI, Apple, or cybersecurity—and I’ll be back tomorrow with the next round of updates.

Google rents SpaceX AI GPUs & Apple’s Core AI and Siri - Tech News (Jun 9, 2026)

Our Sponsors

Today's Tech News Topics