SoftBank, ByteDance Pour $150B Into AI in One Day

SoftBank, ByteDance Commit Over $150 Billion to AI Infrastructure in 24-Hour Spending Blitz

Japan’s SoftBank pledged €75 billion to French data centers, China’s ByteDance announced $70 billion in 2026 compute spending, and Nvidia deepened its CoreWeave bet — as Anthropic took the U.S. Pentagon to court over AI ethics.

SoftBank Group announced the largest AI infrastructure deal in European history on Saturday — a commitment of up to €75 billion to build 5 gigawatts of AI data center capacity across France — as ByteDance simultaneously disclosed plans to spend $70 billion on AI compute in 2026 alone, pushing the total of new commitments disclosed in a single 24-hour period past $150 billion.

SoftBank Picks France for Its Biggest European Infrastructure Push

The deal — announced at the 2026 Choose France Summit and confirmed by SoftBank Group in a press release — centres on building AI data center campuses across three sites in the Hauts-de-France region: Dunkirk (Loon-Plage), Bosquel, and Bouchain. The Bouchain location will be built on a former EDF power plant.

Phase 1, worth €45 billion, is scheduled to deliver 3.1 gigawatts of capacity by 2031. SoftBank is partnering with French state-owned utility EDF for nuclear power supply, with Schneider Electric to manufacture data center components at the Port of Dunkirk, and with French AI startup Sesterce to develop what it calls an “AI factory” at Bosquel.

“This investment reflects our long-term confidence in France as a hub for artificial intelligence infrastructure in Europe.”

— Masayoshi Son, Founder and CEO, SoftBank Group, at the Choose France Summit, May 2026

France’s nuclear grid, which produces electricity at among the lowest carbon intensities in Europe, was described by both SoftBank and EDF as central to the site selection. Under the EU AI Act, energy transparency requirements for large AI deployments are expected to tighten through 2027, making carbon-efficient compute an increasingly material financial consideration.

The investment is expected to create tens of thousands of jobs in engineering, construction, and data center operations across the region, according to the French government’s official summary of the agreement.

ByteDance Eyes $70 Billion in AI Compute Spending for 2026

ByteDance, the Beijing-based parent company of TikTok and Douyin, is planning to spend approximately $70 billion on AI infrastructure during 2026, according to people familiar with the matter cited by multiple technology publications. If confirmed, the figure would represent a single-year capital expenditure larger than the entire annual GDP of several European nations.

ByteDance declined to confirm the specific figure. The company serves approximately 1.7 billion monthly active users across its platforms and runs one of the world’s most complex real-time AI recommendation systems, which processes billions of content-matching decisions per day.

People familiar with the plans said ByteDance is prioritising data center capacity outside of U.S.-controlled cloud providers — specifically reducing its dependence on Amazon Web Services, Microsoft Azure, and Google Cloud — citing concerns about geopolitical exposure and long-term data sovereignty. The company’s Western operations have operated under sustained regulatory scrutiny in Washington.

Nvidia Raises CoreWeave Stake to $3.65 Billion

Nvidia increased its stake in CoreWeave (NASDAQ: CRWV) by 95 percent during the first quarter of 2026, bringing the total value of its position to more than $3.65 billion — approximately 9 percent of the company — according to Nvidia’s quarterly investor filings.

CoreWeave is an AI-native cloud provider that designs its infrastructure specifically for GPU-intensive workloads, unlike traditional hyperscalers that added GPU support to general-purpose cloud architecture. Nvidia has designated CoreWeave as an early deployer of its next-generation hardware, including the Rubin GPU platform, Vera CPUs, and BlueField storage systems. CoreWeave, in turn, has committed to building more than 5 gigawatts of AI data center capacity by 2030.

Nvidia’s investment is part of a broader $3.8 billion in AI portfolio activity during Q1 2026. The company is also assisting CoreWeave in identifying land and power supply sites for new data centers and has agreed to market CoreWeave’s infrastructure to its own cloud partners and enterprise customers, according to Nvidia’s investor disclosures.

Anthropic Sues Pentagon After Being Labelled a National Security Risk

In a separate but equally consequential development, AI safety company Anthropic has filed lawsuits in San Francisco and Washington, D.C., against the U.S. Department of Defense, challenging a “supply-chain risk” designation that has barred defence contractors from working with the company.

The designation — typically reserved for foreign adversaries such as Huawei — was applied to Anthropic after contract negotiations collapsed earlier this year. The company had insisted that any government contract include explicit prohibitions on two uses of its Claude AI models: mass domestic surveillance of U.S. citizens, and fully autonomous lethal weapons systems operating without meaningful human oversight.

The Department of Defense rejected those conditions, demanding unrestricted access for “all lawful purposes.” Following the breakdown, President Trump publicly directed all federal agencies to cease using Anthropic technology, and Defense Secretary Pete Hegseth issued directives to stop commercial activity with the company.

“We cannot in good conscience remove guardrails that exist to prevent the deployment of AI in autonomous lethal systems without human oversight. These are not negotiating positions — they are ethical obligations.”

— Anthropic spokesperson, in a statement to the press, February 2026

OpenAI subsequently signed its own Pentagon agreements. The company stated that its contracts include prohibitions on mass domestic surveillance and requirements for “human responsibility in the use of force,” though the specific terms differ from what Anthropic had proposed, and the full scope of OpenAI’s commitments has not been made public.

By April 2026, signs of a thaw emerged. Anthropic CEO Dario Amodei held what were described as “productive” meetings with White House Chief of Staff Susie Wiles and Vice President JD Vance. President Trump said publicly that a future agreement was possible, noting the technology could be of “great use” to the government. No settlement has been announced as of June 1, 2026.

What Happens Next

SoftBank’s Phase 1 construction in Hauts-de-France is expected to begin within months, pending final permits from local authorities and formal agreements with EDF. The 3.1 GW first phase is scheduled to be operational by 2031.

ByteDance’s infrastructure spending plans for 2026 are subject to board approval and may be adjusted based on macroeconomic conditions and regulatory developments in key markets, according to people familiar with the matter.

CoreWeave is expected to deploy its first Rubin-architecture systems — Nvidia’s next-generation GPU platform — during the second half of 2026. Nvidia has not publicly disclosed a timeline for reducing or increasing its CoreWeave position.

Anthropic’s lawsuits are expected to proceed through the federal court system over the coming months. Legal analysts cited by The Washington Post said the outcome could set binding precedent for how U.S. courts treat government-imposed use constraints on domestic AI companies — a question that has no established legal framework.