Five stories worth your attention this week: a regulatory order that cracks open Android’s AI layer, a compute deal that hints at where Anthropic’s capacity is headed next, a chip startup racing to a $20 billion valuation, a fresh batch of exploits security teams need to patch now, and a layoffs report that puts a number on how much of 2026’s job cuts are actually AI-driven.
1. EU orders Google to open Android to rival AI assistants
On July 16, the European Commission issued two binding orders under the Digital Markets Act. Google must let rival AI assistants use the same system-level Android integrations it currently reserves for Gemini, letting users summon a competing assistant with a voice command the way they’d say “Hey Google” today. Google also has to start sharing anonymized search query and ranking data with eligible competitors starting in January 2027, with full Android interoperability due by July 2027.
Why it matters: if you build on Android, or you’re evaluating AI assistant integrations for a product, the competitive landscape you’re planning around is about to shift. This is one of the more consequential regulatory actions against an AI platform this year, and the phased timeline (2027 for both pieces) gives teams real runway to plan for it rather than react overnight.
Source: Tech Times
2. Anthropic in early talks to lease compute from Meta
Anthropic is reportedly in preliminary talks to lease AI computing power from Meta, according to a July 17 report. This follows a similar arrangement Anthropic already struck with SpaceX to use spare capacity at its Colossus data center to expand serving capacity for paid subscribers.
Why it matters: compute is still the binding constraint for every frontier lab, and Anthropic diversifying its supply across multiple non-traditional partners (SpaceX, and now potentially Meta) is a signal about how tight capacity really is heading into the second half of 2026. For teams building on Claude, this is worth watching as a leading indicator of rate limits and capacity easing (or not) in coming months.
Source: CNBC
3. AI chip startup Etched targets a $20 billion valuation, in two rounds at once
Etched, a startup building chips purpose-built for transformer inference, is reportedly in talks to raise at close to $20 billion, roughly quadrupling its prior valuation, in a round led by existing investor Jane Street. At the same time, it’s running a separate raise led by Sequoia Capital at a lower $10 billion valuation. Neither deal has closed.
Why it matters: running two funding rounds at sharply different valuations in the same window is becoming a pattern among AI infrastructure startups with strong demand, and it’s a useful read on how investors are pricing the inference hardware race against Nvidia. If your team is evaluating inference hardware options beyond Nvidia over the next 12 to 18 months, Etched is now a name to track.
Source: Investing.com
4. This week’s patch list: SonicWall zero-days, an actively exploited ColdFusion flaw, and malicious npm packages
Security researchers disclosed several active threats this week that touch AI and DevOps tooling directly:
- Volexity is tracking a threat actor (UTA0533) exploiting SonicWall Secure Mobile Access 1000 series VPN appliances as zero-days since June 22.
- CISA warned of active exploitation of a critical Adobe ColdFusion flaw, CVE-2026-48282, urging immediate patching across federal and enterprise systems.
- Researchers found seven malicious npm packages targeting the Vite frontend build tooling ecosystem, an expansion of the “ChainVeil” campaign that uses blockchain-based command-and-control infrastructure to stay resilient against takedowns.
Why it matters: two of these three sit directly in common DevOps and frontend tool chains (Vite, ColdFusion), and blockchain-backed C2 is a harder pattern to kill than a standard domain takedown. If your team runs Vite anywhere in its build pipeline, this is worth an audit this week, not next sprint.
Source: eSecurity Planet
5. Report: AI now the top cited reason behind 165,000-plus layoffs in 2026
A report published July 18 found that more than 35 major companies have cited AI adoption as a driver of layoffs so far in 2026, with over 165,000 employees affected across tech, retail, and finance. Oracle, Amazon, Meta, IBM, and Atlassian all appear among the larger cuts, often alongside heavy simultaneous investment in AI infrastructure.
Why it matters: worth holding two things at once here. Companies are citing AI as the reason publicly, but separate analysis (Gartner’s, among others) attributes only a small fraction of actual job losses directly to AI productivity gains, meaning broader economic pressure is doing more of the work than the headlines suggest. For anyone managing a team through this, the distinction between “AI caused this” and “AI was the stated reason for this” matters when you’re planning your own headcount conversations.
Source: ECIKS
The takeaway
The through-line this week is capacity and control: who controls the compute (Anthropic and Meta), who controls the platform (the EU forcing Google’s hand), who controls the hardware (Etched vs. Nvidia), and who’s exploiting the gaps in your tool chain while everyone’s attention is on the model race. Patch your Vite dependencies, and keep an eye on where your team’s AI compute is actually coming from.
For more on building resilient, secure automation on top of these tools, see tha-shed’s Bash Mastery for Cybersecurity course.
