AI News Daily

Issue 60619 · Jun 19, 2026 · 8 stories

Get this in your inbox every morning

Subscribe for the daily AI briefing with curated context and summaries.

Subscribe free
The chip export control saga just got a new twist — the US government is claiming one of ASML's crown jewel EUV lithography machines may have secretly ended up in China, while ASML is flatly saying "no way, we track every single one." Meanwhile, OpenAI is loading up for its IPO by snagging Transformer co-inventor Noam Shazeer from Google DeepMind (yes, *that* Noam Shazeer), and SpaceX is floating the wild idea of putting AI data centers in orbit — though physicists have some thoughts about that. It's a packed day, so let's dive in.

Business, Deals & Funding

TechCrunch AI

The US says ASML’s top chip tool may be in China. ASML says it isn’t

The US says ASML’s top chip tool may be in China. ASML says it isn’t

The US Commerce Department, through Secretary Howard Lutnick, has raised concerns with ASML that one of its extreme ultraviolet (EUV) lithography machines — the only tools capable of printing the most advanced semiconductor patterns — may have ended up in China, which would violate export controls in place since the first Trump administration. US officials claim evidence of EUV-related components and transport equipment shipped to China but have declined to share this evidence with Bloomberg or ASML itself. ASML firmly denies any EUV machine exists or has ever existed in China, with CEO Christophe Fouquet stating the company tracks every machine shipped and maintains strict internal firewalls preventing China-based staff from accessing EUV technology. ASML, Europe's most valuable public company at roughly $700 billion market cap, holds a complete monopoly on EUV lithography tools essent…

Why it matters

This story has significant geopolitical and technological implications. The US claim without shared evidence is concerning — it could represent legitimate intelligence, but it also fits a pattern of using unsubstantiated pressure to gain leverage over allied companies in trade negotiations. ASML's defense is credible: the company has enormous commercial incentive to comply with export controls, as losing its export license would be existentially threatening. The machines are extraordinarily com…

The Verge AI

Barret Zoph is out at OpenAI again after just five months

Barret Zoph is out at OpenAI again after just five months

Barret Zoph has left OpenAI for the second time, departing just five months after returning to the company in January 2026. Zoph had previously left OpenAI to co-found Thinking Machines Lab with former OpenAI CTO Mira Murati, where he served as CTO. Upon his return, he was given the role of head of enterprise AI sales, a strategically important position as OpenAI focused on enterprise revenue ahead of its planned IPO. OpenAI confirmed his departure, and Zoph posted a goodbye message in the company's Slack channels.

Why it matters

This revolving-door departure raises questions about internal stability at OpenAI during a critical period as the company prepares for an IPO. The fact that Zoph was placed in charge of enterprise sales — described as a key revenue driver — and then left after just five months suggests either misalignment on strategy, cultural friction, or possibly a better opportunity elsewhere. It also highlights the broader pattern of high-profile talent churn at OpenAI, which has seen numerous senior depart…

Science Daily

SpaceX wants to build AI data centers in space. Will it work?

SpaceX wants to build AI data centers in space. Will it work?

The article examines the growing interest in building AI data centers in space, particularly by SpaceX, as AI-driven computing demand surges. Proponents argue that orbital facilities could leverage abundant solar energy and bypass Earth-based constraints like land use, water consumption, and local power grid limitations, while also avoiding community backlash. However, engineering professors who study data center design and space systems highlight enormous challenges: Earth-based data centers already require massive electrical power, sophisticated cooling systems, and extensive physical infrastructure. In space, these challenges are amplified—radiation damages electronics, heat dissipation is extremely difficult in the vacuum of space (where there is no air for convective cooling), repairs are extraordinarily expensive, and launch costs remain significant for every pound sent to orbit.…

Why it matters

This is a fascinating but deeply challenging concept that currently sits more in the realm of aspirational engineering than near-term practicality. The fundamental physics problems are daunting: heat rejection in space relies on radiation alone (no convective or evaporative cooling), which requires enormous radiator surfaces and fundamentally limits computing density. Radiation hardening electronics at data center scale would be extraordinarily expensive and would likely impose significant perf…

TechCrunch AI

Source: Elastic agrees to buy CRV-backed DeductiveAI for up to $85M

Source: Elastic agrees to buy CRV-backed DeductiveAI for up to $85M

Elastic has agreed to acquire DeductiveAI, a CRV-backed AI startup focused on catching and resolving software bugs, for up to $85 million. Founded in 2023, DeductiveAI raised a $7.5 million seed round last November at a $33 million valuation, led by CRV with participation from Databricks Ventures, Thomvest Ventures, and PrimeSet. The startup operates in the AI site reliability engineering (AI SRE) space, which is growing due to the massive influx of AI-written code. Elastic plans to integrate Deductive's AI technology into its observability platform to enable automatic performance monitoring and real-time system failure resolution. Despite reaching roughly $1 million in ARR, Deductive's growth reportedly lagged behind competitor Resolve AI, which was last valued at $1.5 billion. Deductive was co-founded by Rakesh Kothari (former VP of engineering at ThoughtSpot) and Sameer Agarwal (form…

Why it matters

This acquisition represents a solid outcome for DeductiveAI's investors, delivering roughly a 2.5x return on a $33 million valuation in under a year since coming out of stealth, though the up-to-$85M price tag suggests a significant portion may be contingent on milestones. The deal highlights the intense demand from established tech companies to acquire AI-native capabilities rather than build them internally. However, the comparison to Resolve AI's $1.5 billion valuation is telling — it sugges…

Ars Technica AI

Microsoft discovers new lightweight backdoor that steals cryptocurrency

Microsoft discovers new lightweight backdoor that steals cryptocurrency

Microsoft has identified a new self-propagating malware called Crypto Clipper that spreads via USB drives and targets cryptocurrency credentials. The worm monitors clipboard contents for wallet addresses and seed phrases, takes screenshots for context, and exfiltrates stolen data through Tor using a SOCKS5 proxy. It spreads through .lnk files on USB drives, disguising itself by naming files similarly to existing ones. The malware also replaces cryptocurrency addresses in the clipboard with attacker-controlled wallet addresses to divert payments. Microsoft describes it as a lightweight backdoor that combines clipboard targeting, screenshot capture, Tor-routed command-and-control communication, and remote code execution capabilities. Microsoft Defender detects it as Trojan:Win32/CryptoBandits.A.

Why it matters

This is a well-crafted piece of malware that demonstrates how relatively simple, script-based tools can be highly effective when combined with smart operational security choices like Tor routing. The USB propagation vector is old-school but still effective, especially in environments where removable media is commonly shared. The clipboard monitoring and address replacement technique is particularly insidious because users rarely double-check pasted wallet addresses character by character. The c…

TechCrunch AI

AI inference startup Baseten reportedly raising $1.5B months after its last mega-round

AI inference startup Baseten reportedly raising $1.5B months after its last mega-round

AI inference startup Baseten is reportedly close to raising $1.5 billion at a $13 billion valuation, just five months after its $300 million Series E at a $5 billion valuation. The round, co-led by Spark Capital, Sands Capital, Altimeter Capital, and Wellington Management, uses a split-priced structure where some investors enter at $13 billion and others at $11 billion. Founded in 2019, Baseten handles AI inference by routing requests to optimal models, including cost-effective open source alternatives, and is part of what has been called the 'inference gold rush.'

Why it matters

The speed and scale of Baseten's fundraising is remarkable — a 160% valuation jump in five months — but the split-priced round structure is a yellow flag that suggests the headline valuation may be somewhat inflated. That said, the inference layer is genuinely becoming a critical and lucrative part of the AI stack. As model commoditization accelerates and open source alternatives proliferate, companies that can intelligently route inference requests to the best-for-task model at the lowest cost…

TechCrunch AI

Snap spins off AI video team into new company, Dotmo, due to costs

Snap spins off AI video team into new company, Dotmo, due to costs

Snap is spinning off its internal generative AI video team into a new separate company called Dotmo, which will focus on developing AI models for interactive gaming experiences. Snap cited high costs of conducting such work internally as a key reason. CTO Bobby Murphy will serve as lead investor with a significant personal stake while remaining at Snap full-time. Snap will receive a large equity stake in Dotmo and provide a technology license. The Dotmo team will consist of current Snap employees leaving the company. This is Snap's second major spinoff in 2026, following the earlier separation of its Specs smart glasses division, and comes after the company cut approximately 1,000 jobs earlier in the year.

Why it matters

This spinoff appears to be a pragmatic cost-cutting move disguised as strategic flexibility. Snap is essentially offloading expensive AI R&D while maintaining upside through equity — a clever financial maneuver. However, it also signals that Snap may be struggling to justify the costs of ambitious AI projects within its core business. The pattern of multiple spinoffs and layoffs in 2026 suggests a company under significant financial pressure, trying to streamline around its core Snapchat produc…

TechCrunch AI

OpenAI is bringing on some big guns in the lead-up to its IPO

OpenAI is bringing on some big guns in the lead-up to its IPO

OpenAI is making major hires ahead of its IPO, recruiting Google DeepMind's Noam Shazeer — a co-author of the foundational 'Attention Is All You Need' Transformer paper and former Character AI founder — and Dean Ball, a former Trump White House AI policy official. Shazeer left Google where he had co-led Gemini development, while Ball will lead a new 'Strategic Futures' team at OpenAI focused on frontier AI policy, catastrophic risk, recursive self-improvement, and government relations. Ball's hire is particularly notable given the current political landscape: Anthropic recently faced a Trump-ordered export control ban on its latest models, highlighting the strategic importance of government relationships for AI companies. Ball emphasized that internal governance at AI labs will become increasingly central to AI's future.

Why it matters

This is a significant power consolidation move by OpenAI on two fronts. Landing Shazeer is a genuine technical coup — he's one of the most important figures in modern AI research, and his departure from Google DeepMind is a real blow to that organization. The Dean Ball hire is arguably even more strategically important and considerably more concerning. The juxtaposition with Anthropic's export ban troubles makes the political dynamics impossible to ignore: OpenAI is essentially building a revol…

From X/Twitter

From Reddit/HN/YC

Never miss the next issue

Read on the web or get tomorrow's issue delivered directly by email.

Join AI Newsy