AI IPO wave builds as SoftBank plants €75bn in France
OpenAI and SpaceX race toward public listings as SoftBank commits €75bn to French AI infrastructure and new governance rules take shape.
By BINA Editorial
Today's brief is anchored by money at a scale that reshapes geography: a Japanese conglomerate is remaking northern France into one of the world's largest AI computing zones, two tech giants are preparing imminent public listings, and a thirty-day security scan has already surfaced software flaws hidden for decades.
SoftBank pledges up to €75bn to build AI data centres in France
At the 2026 Choose France summit, SoftBank Group committed to develop and operate up to 5 gigawatts of AI data-centre capacity across France, representing an investment of up to €75 billion — its largest AI infrastructure commitment in Europe. The first phase targets 3.1 GW across three sites in the Hauts-de-France region, including Dunkirk (Loon-Plage), Bosquel, and Bouchain, with delivery expected by 2031. SoftBank will collaborate with Schneider Electric to establish a manufacturing cluster at the Port of Dunkirk for data-centre components and power modules, tying large-scale computing directly to industrial employment in the region.
OpenAI engages Citigroup and JPMorgan ahead of September IPO
OpenAI has entered discussions with Citigroup and JPMorgan Chase about joining Goldman Sachs and Morgan Stanley on the underwriting team for its anticipated initial public offering. A confidential S-1 filing is expected within weeks, with a public listing targeted for September 2026. The company, which has raised $186 billion in private markets, is expected to seek a valuation of up to $1 trillion — a listing that would rank among the largest in Wall Street history if it proceeds on schedule.
SpaceX files for IPO, disclosing $4.3bn in xAI-related losses
SpaceX filed an S-1 for an initial public offering that analysts project could value the company between $1 trillion and $1.75 trillion. The filing disclosed $4.3 billion in net losses in the first quarter of 2026, largely driven by xAI, Elon Musk's artificial intelligence unit. Governance terms in the prospectus attracted scrutiny: shareholders would have limited influence over Musk, who retains concentrated control, with performance-based compensation tied to milestones including human settlement on Mars. The overlap between SpaceX, xAI, and other Musk ventures has drawn concern from institutional investors.
Anthropic releases Opus 4.8 and finds 10,000 software flaws in 30 days
Anthropic launched Claude Opus 4.8 on 28 May 2026 — the shortest interval between consecutive Opus releases. The model is four times less likely to produce flawed outputs without flagging them than its predecessor, Opus 4.7, and introduces a "high-effort" mode for complex reasoning tasks along with Dynamic Workflows, which allow Claude to orchestrate multiple AI subagents working in parallel. Separately, Anthropic's defensive cybersecurity initiative, Project Glasswing — powered by the unreleased Claude Mythos Preview model — identified more than 10,000 high- or critical-severity software vulnerabilities in its first 30 days, including a 27-year-old flaw in OpenBSD and a 16-year-old flaw in the widely used media library FFmpeg. The scale of the findings has shifted attention from locating bugs to the harder problem of fixing them quickly enough.
Huawei proposes Tau Scaling Law to carry chips beyond Moore's Law
At the 2026 IEEE International Symposium on Circuits and Systems in Shanghai, Huawei unveiled the "Tau Scaling Law," a new framework for semiconductor advancement intended to carry the industry beyond the physical limits of Moore's Law. Where Moore's Law focused on shrinking transistors, the Tau framework optimises signal transmission times across multiple levels simultaneously — semiconductor devices, circuits, chips, and full computing systems — to improve speed and energy efficiency. Huawei's roadmap targets 1.4-nanometre nodes by 2031, signalling that one of the world's largest chipmakers is formally moving away from transistor count as its primary performance yardstick.
Cognition raises $1bn as Devin writes 89% of its own engineers' code
Cognition, the company behind Devin — an AI software-engineering agent — closed a $1 billion Series D round at a valuation of $26 billion, up from $10.2 billion in September 2025. The company disclosed a striking internal figure: Devin now writes 89% of the code committed by Cognition's own engineers, making the startup a living demonstration of its own thesis. The funding will support further development of Devin and infrastructure scaling.
OpenAI publishes Frontier Governance Framework aligned with EU and California law
OpenAI released its Frontier Governance Framework on 29 May 2026, a public document mapping its internal safety and security practices to the requirements of California's Transparency in Frontier AI Act and the EU AI Act's Code of Practice for General Purpose AI. The framework covers specific risk domains: cyber-offence capabilities, chemical and biological threats, and harmful manipulation. Separately, the OpenAI Foundation announced $250 million in grants to study how AI is reshaping economic measurement, noting that traditional indicators such as wages and employment may no longer capture the full picture of human welfare in an AI-augmented economy.