Business Technology News Roundup: Jan 30, 2026

The biggest story of  the week, and potentially the decade, is that SpaceX and xAI are in formal  merger talks. Elon Musk effectively confirmed the rumors on X with a  characteristically blunt "Yeah" in response to reports about the  deal.

SpaceX (currently  valued at $800 billion) and xAI (valued at $230 billion) are exploring a  stock swap that would bring them under a single corporate umbrella. Nevada  state filings from January 21 show the creation of two new entities managed  by SpaceX CFO Bret Johnsen, which experts believe are the legal  "buckets" for this transaction.

This isn't just a  financial maneuver; it’s a bid to create a "Dyson Swarm" company.  The vision is to use SpaceX’s Starlink satellite constellation as a global,  space-based supercomputer to host xAI’s "Grok" intelligence. By  moving the "brain" of AI into orbit, Musk aims to bypass  terrestrial power constraints and provide low-latency AI services anywhere on  Earth. With a SpaceX IPO rumored for later in 2026, this merger would create  a trillion-dollar powerhouse that is virtually impossible for competitors to  replicate.

While OpenAI usually  hogs the headlines, Anthropic just proved that the hunger for  "safe" AI is bigger than ever. They reportedly doubled their latest  funding target mid-round, closing at a $350 billion valuation.

This massive $20  billion injection of cash was led by Sequoia Capital, with major  participation from Singapore’s sovereign wealth fund (GIC) and Coatue. This  is a jaw-dropping jump from their $183 billion valuation just four months  ago.

What’s driving this  frenzy? It’s two-fold: the successful commercialization of "Claude  Code", a tool that has become the gold standard for AI-assisted  programming, and the company's reputation for "Constitutional AI"  (AI with a built-in moral compass). Investors are flocking to Anthropic as  the "stable" alternative to OpenAI’s more aggressive,  research-first culture. This valuation places Anthropic among the top 20 most  valuable entities on the planet, private or public, signaling that the  "AI bubble" isn't bursting, it’s just getting more expensive to  join the race.

On the security  front, we saw a terrifying milestone. A botnet known as Aisuru (or Kimwolf)  launched a record-breaking DDoS attack that peaked at 31.4 terabits per  second.

Cloudflare, which  mitigated the attack, named the campaign "The Night Before  Christmas." The attack reached a hyper-volumetric rate of 200 million  requests per second, primarily targeting telecommunications providers.

The most alarming  part? The power didn't just come from compromised industrial servers. A  massive portion of the traffic was traced back to Android TVs and streaming  boxes. This highlights a growing trend in 2026: "Living Room  Warfare." Hackers are exploiting the lax security of smart home devices  which are rarely updated and always connected to build digital armies. If you  haven't checked your router or smart TV settings lately, this is your wake-up  call that your Netflix box might be a soldier in a global cyberattack.

In a move straight  out of science fiction, Alibaba’s Qwen-3 model became the world’s first  general-purpose AI to run full inference tasks entirely in orbit.

Working with the  aerospace startup Adaspace, Alibaba successfully deployed the model to an  orbital computing center consisting of a 12-satellite constellation. In a  successful test on January 27, queries were sent from Earth, processed  entirely by the AI on the satellite, and the results were returned in less  than two minutes.

This is a  game-changer. Traditionally, satellites are "dumb" sensors that  send raw data back to Earth for humans (or ground-based AI) to analyze. By  putting the "brain" on the satellite itself, Alibaba can process  Earth-observation imagery or disaster response data in real-time. It’s the  first step toward a 2,800-satellite "Star-Compute" network planned  for 2035, which could provide AI services to autonomous drones and robots  globally without ever needing a ground-based server.

If you’re a  developer, it’s time to audit your plugins. Security researchers flagged a  campaign dubbed "MaliciousCorgi," involving two popular VS Code  extensions that have been harvesting source code from over 1.5 million users.

The extensions,  "ChatGPT - 中文版" and "Chat-Moss," appear to be legitimate,  functional AI assistants. They offer great autocomplete and debugging  features, which is exactly how they stayed undetected for so long. However,  hidden under the hood is a spyware infrastructure that triggers every time  you open a file.

Unlike normal  extensions that only read the snippet you're working on, MaliciousCorgi  encodes entire files and sends them to servers in China. It even features a  "remote harvest" command where the attacker can trigger the  extension to steal 50 files from your workspace at once without any visual  indication. It’s a sobering reminder that as we rush to add AI  "helpers" to our workflow, we are essentially inviting a stranger  to sit behind us and watch every keystroke. If you use VS Code, check your  extensions list immediately.