Some weeks in tech feel like noise. Press releases dressed up as announcements, product updates nobody asked for, and executive quotes that say nothing in the most polished way possible.
This week was not one of those weeks.
Between Google exploring data centers in literal outer space, the first documented case of hackers using AI to discover and weaponize a software vulnerability, and the four biggest tech companies in the world committing to a combined $725 billion in spending this year alone — there is a lot here that deserves more than a headline.
Here is what actually happened, why it matters, and what it means for anyone who pays attention to where technology is heading.
1. Google Is Talking to SpaceX About Putting Data Centers in Orbit
This one sounds like a technology joke until you read the details. Google is in serious discussions with SpaceX about launching data centers into orbit as a way to meet the exploding computing demands of its AI infrastructure.
The pitch from SpaceX is that orbital data centers could be cheaper to operate than ground-based ones when you factor in land costs, power infrastructure, and the regulatory battles that come with building massive facilities in populated areas. Elon Musk has been public about his belief that putting compute in space avoids the NIMBY problems that have slowed data center construction in parts of the US and Europe.
Google already invested around $900 million in SpaceX back in 2015, so the relationship is not new. What is new is the specific application being discussed and the urgency behind it. The company’s Project Suncatcher aims to have prototype satellites deployed by 2027, which means this is not a ten-year vision being floated at a conference — it is an active engineering project with a timeline.
The honest context here is that data centers have a power problem that is becoming impossible to ignore. AI workloads consume enormous amounts of electricity. Finding land near reliable power sources with cooling infrastructure and political approval has become genuinely difficult in many parts of the world. Space removes most of those constraints, at the cost of a different set of engineering challenges that SpaceX has more experience with than virtually anyone else on earth.
Whether orbital data centers become mainstream infrastructure or a fascinating experiment is genuinely unclear. But the fact that Google is treating this as a serious option rather than a thought experiment says something significant about how acute the compute infrastructure problem has become.
2. Hackers Used AI to Find a Zero-Day Vulnerability — and It Worked
Google’s security team reported this week the first documented case of criminal actors using AI to discover a zero-day vulnerability and develop a working exploit from it. Google caught the attempt and blocked it before real damage was done, but the significance of the report goes beyond this specific incident.
A zero-day is a software vulnerability that is unknown to the people responsible for fixing it. Finding them has traditionally required significant technical skill, time, and often a degree of luck. Security researchers spend careers looking for them. Criminal groups with serious resources find a handful per year.
AI has changed that calculation in a way that the security industry has been warning about for two years. The tools now exist to automate the process of scanning code for patterns associated with vulnerabilities, generating test cases, and iterating on exploits at a speed and scale that human researchers cannot match. What this week’s report confirms is that criminal groups are not just theoretically capable of using these tools — they are actively doing it.
The defensive side of this equation is also changing. AI-powered security tools can scan for vulnerabilities faster than human teams. The honest question nobody has a clean answer to yet is whether the offense or the defense gets more leverage from these capabilities. The early evidence from this week suggests the gap between attacker sophistication and defender readiness is worth taking seriously.
3. Big Tech Is Spending $725 Billion This Year — While Cutting Jobs
Meta, Amazon, Microsoft, and Alphabet have collectively committed to approximately $725 billion in capital expenditure for 2026. That number represents an increase of more than 75 percent compared to last year, and almost all of it is going toward the same category: data centers, custom chips, GPUs, and AI model infrastructure.
At the same time, the same companies are reducing headcount in ways that are not small. Meta plans to cut 8,000 employees in May. Amazon has eliminated roughly 30,000 roles in recent months. Microsoft has offered voluntary exits to around 125,000 employees. GitLab announced layoffs this week and described the move as freeing up resources for future priorities rather than cost-cutting, which is technically a different thing but lands the same way for the people affected.
The pattern here is not complicated once you see it clearly. These companies have decided that the next phase of competition is won or lost at the infrastructure layer, not the headcount layer. Paying for compute is a better bet, in their view, than paying for the people who previously did the work that AI systems can now approximate.
Whether this reallocation creates more value than it destroys is the open question. AI infrastructure spending creates jobs in construction, hardware manufacturing, and chip design. It eliminates them in software, operations, and administrative functions. The net effect on employment, and on the quality of jobs that remain, will take years to measure accurately. The decisions are being made now.
4. Anduril Just Raised $5 Billion. Defense Tech Is No Longer a Fringe Category.
Anduril Industries — a defense technology startup founded in 2017 by Palmer Luckey, the creator of the Oculus VR headset — raised $5 billion this week at a valuation of $61 billion. The round was led by Thrive Capital and Andreessen Horowitz.
The company builds drones, autonomous surveillance systems, and AI-powered command-and-control software for military applications. Its Lattice platform, which uses AI to coordinate autonomous systems on the battlefield, is at the center of a $20 billion US Army contract that gave the company its most significant proof of institutional confidence.
What makes this funding round notable beyond the numbers is what it signals about the broader venture capital market. Defense tech was, for most of Silicon Valley’s history, a category that prominent investors avoided or treated as a niche. The combination of rising global military spending, the demonstrated effectiveness of drone warfare in recent conflicts, and the maturity of AI autonomy systems has changed that calculation significantly.
Startups that can move faster than traditional defense contractors and build software-first systems are increasingly winning major government contracts. Anduril’s raise is the clearest signal yet that venture capital has decided this category is real and that the companies building in it can reach the kind of scale that justifies institutional-sized bets.
5. OpenAI Is Being Sued — and Sam Altman Is Taking the Stand
The legal case that Elon Musk filed against OpenAI moved into active courtroom proceedings this week, with Sam Altman scheduled to testify. The case centers on Musk’s argument that OpenAI’s transition from a nonprofit to a for-profit structure violated the founding agreements that led him to contribute early funding.
The substance of the dispute is genuinely complicated. OpenAI was founded with an explicit mission to develop artificial general intelligence for the benefit of humanity rather than shareholders. As the company has raised billions in commercial funding and built a business generating significant revenue from enterprise customers and consumers, the distance between that original mission and its current operating model has become harder to ignore.
Musk’s motivations in bringing the case are not entirely clean — he has his own AI company in xAI and a clear competitive interest in challenging OpenAI’s position. Courts tend to be aware of this kind of context.
The more significant outcome of this case, regardless of how the specific legal arguments resolve, is likely to be the precedent it sets for how AI companies can evolve their governance structures after raising money under nonprofit or mission-driven frameworks. Several other organizations in the AI space are watching closely because variations of the same tension exist in their own structures.
6. Apple Released iOS 26.5 — With Encrypted RCS Messaging
Apple shipped iOS 26.5 this week with a feature that sounds technical but has practical significance for anyone who sends text messages across platforms: end-to-end encryption for RCS messaging, enabled by default through supported carriers.
RCS is the messaging standard that replaced SMS for most Android users years ago. Apple’s adoption of RCS last year removed the green bubble problem that had made cross-platform messaging awkward. End-to-end encryption in RCS takes it a step further, meaning messages sent between iPhones and Android devices are now encrypted in transit in the same way iMessage conversations between two iPhones have always been.
The practical impact is that a significant category of everyday communication — the cross-platform messages that billions of people send daily — becomes meaningfully more private. Carriers cannot read them. Hackers intercepting network traffic cannot read them. The people sending them can.
This is the kind of update that does not generate the same headlines as a new chip announcement or a redesigned camera app, but its effect on how private ordinary communication is affects more people than almost any other change Apple ships in a given year.
7. Big Tech Layoffs Are Still Happening — and the Reasons Are Changing
The tech layoff story of 2022 and 2023 was largely about companies that had over-hired during the pandemic growth period and were correcting. The layoffs happening now are structurally different, and conflating the two gives a misleading picture of what is actually going on.
The current round of cuts at companies like Meta, Microsoft, Amazon, and GitLab is not primarily about right-sizing headcount after a hiring spree. It is about reorienting what kinds of work the company needs humans to do versus what AI systems can now handle. Customer support, content moderation, certain categories of software development, data labeling, and administrative coordination are all areas where the cost-benefit analysis of human labor versus AI tooling has shifted meaningfully in the last 18 months.
The jobs being created in the same companies tend to be in AI research, infrastructure engineering, and the specialized roles needed to deploy and maintain AI systems at scale. These are real jobs, but they require different skills and there are fewer of them relative to the roles being eliminated.
Nobody should be surprised by this. The question that matters is whether the retraining and transition support available to displaced workers is anywhere close to adequate for the speed at which this transition is happening. The honest answer, looking at what most companies and governments are actually providing, is no.
8. The AI Chip War Between the US and China Just Got More Complicated
Nvidia’s CEO Jensen Huang joined a US government delegation to China this week in a diplomatic trip that reflects how central semiconductor access has become to the relationship between the two countries.
The export restrictions the US has placed on advanced AI chips — specifically Nvidia’s highest-performance GPUs — have been one of the defining features of US-China tech policy for the past two years. The intent is to slow China’s ability to develop and deploy frontier AI systems by limiting access to the compute needed to train and run them.
The practical effect has been complicated. Chinese companies have accelerated domestic chip development faster than many observers expected, though they remain behind at the leading edge. They have also found ways to access restricted chips through third-party routes that have proven difficult to fully close. And the restrictions have pushed Chinese AI companies to develop more efficient training and inference methods that, in some cases, have produced competitive results using less compute than US models require.
DeepSeek, a Chinese AI lab, reportedly entered discussions this week to raise its first external funding at a valuation around $45 billion, with China’s state-backed semiconductor fund involved. The company’s work on models optimized for domestic chips — which are less powerful than Nvidia’s restricted exports — has produced results that surprised the US AI industry and forced a genuine reconsideration of how much the compute advantage actually determines outcome.
The chip war is not going away. It is getting more complex, and the early assumption that restricting hardware access would translate cleanly into restricting AI capability has run into a more complicated reality.
This Week’s Stories at a Glance
| Story | What Happened | Why It Matters |
| Google x SpaceX | Talks on orbital data centers | AI compute problem goes to space |
| AI Zero-Day Hack | First AI-powered exploit confirmed | Defense teams need to move faster now |
| $725B Big Tech Spend | Record capex + simultaneous layoffs | Infrastructure over people is the bet |
| Anduril $5B Raise | $61B valuation, defense AI booming | VC has decided defense tech is real |
| OpenAI Trial | Altman testifying, Musk suing | Precedent for AI company governance |
| iOS 26.5 RCS | Encrypted cross-platform messages | Billions of daily texts now private |
| Tech Layoffs 2.0 | AI reorg, not pandemic correction | Different jobs, fewer of them |
| US-China Chip War | Jensen Huang in Beijing, DeepSeek $45B | Compute restrictions hitting limits |
