Tech News — 2026-06-22#

Story of the Day#

The US government has placed unprecedented export controls on Anthropic’s newest AI models, Mythos and Fable, barring foreign nationals from accessing them over national security concerns. The sudden ban follows intense lobbying by Anthropic itself, whose executives repeatedly warned that the models’ advanced coding capabilities could pose a global cybersecurity threat.

Top Stories#

Valve Prices the Steam Machine at $1,049 · The Verge Valve’s highly anticipated living room PC, the Steam Machine, will launch on June 29 with a staggering $1,049 starting price for the 512GB model. Valve executives attribute the high cost to a brutal RAM and component shortage, noting they are selling the hardware at cost rather than subsidizing it. In a simultaneous move to preserve the open PC ecosystem, Valve announced it is officially opening up SteamOS for users to install on their own desktop hardware.

Microsoft and Chevron plan one of the largest gas-powered data center projects in US · TechCrunch Highlighting the immense energy appetite of artificial intelligence, Microsoft has inked a 20-year power purchase agreement with Chevron to build a natural gas-fired power plant for a massive new data center in West Texas. The Pecos datacenter campus will ultimately require up to 2.67 gigawatts of power, operating “behind the meter” and completely independent of the public grid. This locks in decades of carbon emissions, complicating the tech giant’s aggressive climate pledges.

WhatsApp gets new chief as Meta taps India’s CRED founder Kunal Shah, and invests $900M in startup · TechCrunch Will Cathcart is stepping down from his role as the head of WhatsApp after seven years. In a major shakeup, Meta has appointed Kunal Shah, founder of the Indian fintech startup CRED, to replace him. As part of the leadership transition, Meta is injecting $900 million into CRED for a 20 percent stake, signaling a massive push to integrate robust financial and payment services directly into the messaging app.

Google DeepMind bets $75M on AI’s future in Hollywood with A24 deal · TechCrunch Google’s DeepMind is investing $75 million into A24 to collaboratively develop new AI-powered filmmaking tools. While AI has sparked fierce backlash and lawsuits across Hollywood, A24 insists this non-exclusive partnership is focused on creating workflows like AI-generated storyboards that preserve creative control rather than replacing human production.

AWS Graviton5 Reaches General Availability with 192 Cores and Formally Verified VM Isolation · InfoQ Amazon Web Services officially rolled out its Graviton5 processors, doubling the core count to 192 per chip on TSMC’s 3nm node. AWS is heavily marketing the chips for “agentic AI” workloads, supported by massive deployments from Meta, Uber, and Snowflake. Notably, the instances feature the first formally verified hypervisor isolation in a production cloud environment, offering mathematically proven security guarantees between virtual machines.

Also Worth Knowing#

• Following user outcry, AMD reinstates memory encryption in consumer CPUs (Ars Technica): After silently stripping Transparent Secure Memory Encryption (TSME) from its non-PRO Ryzen 9000-series processors via a firmware update, AMD has promised to restore the physical security feature in July following intense backlash from users.
• TikTok Shows 3x More AI Slop Than YouTube, Report Finds (Slashdot): A new analysis revealed that 59% of videos served to fresh TikTok accounts are AI-generated “slop,” compared to just 21% on YouTube Shorts, with the kids’ category being the most heavily impacted.
• Oracle Cut 21,000 Employees in 2026, Says AI Replaced Some Jobs (Bloomberg): Oracle has quietly reduced its workforce by 21,000 employees over the past twelve months, explicitly acknowledging that artificial intelligence tools have eliminated some of those roles.
• SpaceX inks compute deal with Reflection AI, an open-source AI lab (TechCrunch): SpaceX will provide Reflection AI with immediate access to Nvidia’s GB300 chips in its Memphis datacenter, charging $150 million a month as Elon Musk’s firm pivots toward AI infrastructure.
• Doorbell cam filmed Tesla Autopilot crash that killed woman in her home (Ars Technica): Federal regulators are investigating a fatal crash in Texas where a driver operating a Tesla Model 3 in an automated driving mode veered off the road and struck an elderly woman’s home at high speed.