America shipped. China’s still ahead. NVIDIA released Nemotron 3 Ultra on June 4, a 550-billion-parameter open model with 55 billion active per token that serves over 400 tokens per second. It scores 47.7 on the Artificial Analysis Intelligence Index, topping Gemma 4 31B at 39.2 and gpt-oss-120b at 33.3. But China’s Kimi K2.6 still sits at 53.9. The US leads on open-weight speed and availability; the benchmark crown stays abroad.
The business of AI is maturing on two fronts at once. Quantinuum’s $1.68B IPO priced 28 million shares at $60, above range, for a valuation near $14 billion at the offering, with Honeywell holding about 48.1% voting power. Days earlier, the Claude Partner Network added a three-tier services track; 40,000 firms have applied since its March launch and 10,000-plus consultants are now certified. Public markets and enterprise distribution are forming at the same moment.
Europe is changing tactics. The EU’s sovereignty package, unveiled June 3, pairs a Cloud and AI Development Act with a revised Chips Act. The Commission estimates roughly EUR 120 billion for semiconductors, EUR 200 billion for data centers by 2036, EUR 100 billion for cloud and AI, and EUR 2 billion for open source. The stated goal is cutting reliance on non-EU cloud providers. Brussels is shifting from regulating AI to funding a sovereign stack.
The Pulse
Sam Altman lobbies Congress to fund AI testing, not gate model releases
Fund the referees, skip the permission slips — Silicon Valley’s preferred kind of oversight.
Suno raises another $400M at a $5.4B valuation while the copyright suits pile up
Investors are betting the lawsuits get settled, not lost.
Perplexity tests a local-cloud router to keep AI PC data private
Privacy as a routing decision — the on-device pitch finally gets practical.
That’s the signal.
— The PulseMark Team
Get the Daily Pulse
Sharp analysis on what's actually moving in AI. No hype, no filler, no weekly digest.