Europe and UK Race to Secure Sovereign AI

The UAE left OPEC on May 1, freeing up more than $61 billion a year in potential oil revenue and enabling accelerated energy and AI investments. ADNOC immediately announced $55 billion in faster spending on production, refining and petrochemicals, while state-backed funds like MGX are committing up to $10 billion annually to AI deals and have co-invested in OpenAI and Anthropic. Emirati conglomerate G42 is building a five-gigawatt Stargate campus for OpenAI and, via Khazna, anchors much of the country’s data-center capacity; Microsoft is investing $15.2 billion in UAE data centers. ADNOC’s XRG arm is pursuing U.S. gas assets to secure power for data centers, highlighting how energy supply and sovereign capital are being redeployed to fuel AI infrastructure.

Anthropic announced a major compute partnership with SpaceX/xAI to take over Colossus I capacity immediately, a deal reported by some to be roughly $5B/year for ~300MW that will sharply boost Claude product limits and ramp inference on Colossus within days. The company used its developer event to detail Claude Managed Agents, expanded Claude Code rate limits and higher API caps, and celebrated rapid ARR and usage growth (ANNUALIZED expansion cited ~8,000%). CTO Tom Brown and other leaders framed this as solving compute bottlenecks to scale multi-agent and enterprise services, while Dario Amodei highlighted tiny teams, multiagent systems, and enterprise productivity as strategic focuses. The move reshapes cloud dynamics and intensifies competition with OpenAI/xAI for large-scale model deployment.

Anthropic and OpenAI are doubling down on services as the next major AI revenue frontier, each announcing large, PE-backed service companies to package models into enterprise workflows. Anthropic formed a $1.5B joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs to build Claude-powered, tailored systems for customers; OpenAI’s The Deployment Company has raised roughly $4B at a $10B pre-money valuation with investors like TPG and Bain to commercialize software sales through a PE partnership. The moves reflect a wider industry bet that deploying agents into real-world knowledge work requires integration, IT upgrades, change management, and vertical services—creating space for startups like Tessera and in-house vertical initiatives (finance noted as a key segment). OpenAI also rolled out GPT-5.5 Instant, personalization, real-time infra updates, and expanded developer agent tooling.

Anthropic announced a major compute partnership with SpaceX to use all capacity at SpaceX’s Colossus 1 data center, unlocking more than 300 megawatts (over 220,000 NVIDIA GPUs) within a month and boosting capacity for Claude Pro and Claude Max users. The company also raised usage limits—doubling Claude Code’s five-hour rate limits for paid plans, removing peak-hour reductions for some accounts, and increasing API rate limits for Claude Opus models. Anthropic framed the deal alongside other large-scale capacity agreements with Amazon, Google/Broadcom, Microsoft/NVIDIA, and Fluidstack, and said it trains Claude on AWS Trainium, Google TPUs and NVIDIA GPUs. The firm plans international expansion for regulated-enterprise needs and signaled interest in orbital AI compute with SpaceX.

Anthropic is on track to generate about $10.9 billion in revenue in Q2, according to a source cited by the report. The AI startup, known for developing large language models, has rapidly scaled commercial uptake amid broader industry momentum that includes competitors like OpenAI and chip suppliers such as Nvidia. That revenue projection underscores robust demand for advanced AI services and reinforces investor expectations ahead of possible IPO activity in the sector. It matters because outsized revenue from private AI firms can reshape funding, valuations and talent flows across the cloud, chip and software ecosystems, while intensifying regulatory and competitive scrutiny of dominant models and platform partnerships.

Anthropic is on track to generate about $10.9 billion in revenue in the second quarter, according to a source cited by the report. The AI startup, backed by investors and competing with the likes of OpenAI, has rapidly expanded commercial deployments of its models and services, fueling substantial sales growth ahead of a potential IPO. That surge underscores intense market demand for generative AI and could influence valuations, hiring, and competitive dynamics across the AI sector, while putting pressure on rivals and cloud/infrastructure partners to scale. The figure highlights Anthropic’s rising commercial clout and the broader commercialization boom driving investor interest in AI companies.

Anthropic has reportedly agreed to pay SpaceX roughly $1.25 billion per month for access to the company’s GPU-heavy data center and networking capacity to run large language models. The deal, if accurate, reflects how hyperscale compute and specialized infrastructure are becoming critical bottlenecks for AI firms and highlights vertical partnerships between model developers and infrastructure providers. Anthropic gains dedicated, low-latency connectivity and custom hardware scaling; SpaceX monetizes its Starlink-connected edge and data center footprint. The arrangement matters because it underscores soaring operational costs for advanced AI, could reshape cloud and GPU markets, and may prompt more exclusive infrastructure deals between AI companies and nontraditional providers.

SpaceX disclosed in its S-1 that it will both use its COLOSSUS and COLOSSUS II data centers for proprietary AI like Grok 5 and sell excess compute to third parties. Notably, SpaceX signed Cloud Services Agreements with Anthropic in May 2026 that commit the customer to $1.25 billion per month through May 2029, with capacity ramping in May–June 2026 at a reduced fee; either party can terminate with 90 days’ notice. This deal signals major cloud capacity monetization by a nontraditional infrastructure provider and underscores rising commercial demand for large-scale AI training resources, with implications for cloud competition, AI model scaling, and data center economics. The quote was shared by Simon Willison.

OpenAI and Malta partner to bring ChatGPT Plus to all citizens

Alphabet (Google’s parent) will invest $15 billion to build a new data center campus in New Florence, Missouri, marking one of the state’s largest tech infrastructure projects. The project responds to surging cloud and AI compute demand and includes agreements to secure over 1 GW of new power generation capacity and, with utility Ameren, more than 500 MW of supporting electric capacity. The buildout signals major long-term commitments to regional power and data infrastructure and underlines ongoing cloud providers’ competition for large-scale AI compute and resilient power supply.

The FAA says SpaceX has proposed an ambitious plan to reach 10,000 launches per year within five years, but regulators want to see improved reliability before approving such expansion. The report also notes SpaceX earlier outlined plans to deploy a constellation of one million satellites that would orbit Earth and use solar power to feed AI data centers. The scale of launches and satellite deployments could reshape space-based communications and AI infrastructure, but raises regulatory, safety and operational questions that US agencies need to evaluate. The development matters for satellite internet, cloud/AI compute offloading, and the commercial space industry.

Anthropic told investors it expects to more than double revenue to about $10.9 billion in Q2 2026 and to post an operating profit for the first time, according to the Wall Street Journal. The company’s chatbot Claude and expanded services for small businesses and law firms have driven rapid growth, positioning Anthropic more competitively against OpenAI — which faced simultaneous reports of an impending IPO. However, the WSJ noted Anthropic’s profitability may be short-lived given large upcoming compute costs tied to its operations. The financials were disclosed during a funding round; Anthropic declined further comment.

SpaceX disclosed Cloud Services Agreements in its S-1 showing Anthropic will pay $1.25 billion per month through May 2029 for access to COLOSSUS and COLOSSUS II compute capacity, with ramping in May–June 2026. The deal would make cloud compute rental one of SpaceX’s largest revenue streams, dwarfing Starlink income if fully realized, and raises questions about capital recovery, electricity and financing costs, and long-term viability if Anthropic exits. The filing triggered debate over Anthropic’s prior commitments on measured compute growth and whether they’re now paying premium prices. Analysts and commenters also discussed how much capacity Anthropic actually leased and the implications for SpaceX’s data-center economics.

Anthropic is expanding model hosting to Colossus2 and will use the GB200 accelerator, signaling continued scaling of its infrastructure. The move was reported in a Hacker News thread discussing local environmental and health concerns tied to Colossus datacenter sites — notably emissions from on-site gas turbine generators and regional pollution. Community commenters debated trade-offs between economic investment and local externalities, compared Colossus to other projects like xAI’s sites, and shared investigations and reactions. This matters because Anthropic’s choice of hardware (GB200) and datacenter providers affects model performance, cost, and the broader debate over the environmental footprint of AI infrastructure.

OpenAI is preparing to confidentially file a draft IPO prospectus as soon as this Friday, working with banks including Goldman Sachs and Morgan Stanley, CNBC reports. Valued at over $850 billion by private investors and having raised more than $18 billion, the company has been planning a potential IPO as early as Q4 and says its focus "remains on execution." The move comes amid legal friction with Elon Musk, competition from rivals like Anthropic, and scrutiny over financials and burn rate. A confidential filing would be a key step toward what could be one of the largest public debuts in history and would reshape the AI industry’s public-market landscape.

OpenAI is preparing to confidentially file a draft IPO prospectus as soon as Friday, CNBC reports, working with banks including Goldman Sachs and Morgan Stanley. Valued at over $850 billion by private investors, OpenAI has been planning a potential public offering as early as Q4 and says it regularly evaluates strategic options while focusing on execution. CFO Sarah Friar framed IPO readiness as good corporate hygiene. The move follows rival SpaceX’s confidential SEC filing and comes amid legal friction with Elon Musk and intensifying competition from firms like Anthropic. An IPO would be one of the largest debuts ever and will test OpenAI's financials as it burns cash and scales commercial offerings.

Anthropic is forecast to post its first profitable quarter in 2026 Q2, with estimated revenue of $10.9 billion and operating profit around $559 million, up from $4.8 billion in Q1, according to the Financial Times. The San Francisco AI company’s surge would outpace rivals OpenAI and xAI, both still recording heavy investments and losses as they scale compute and model operations. OpenAI is not expected to reach profitability until about 2030 amid cumulative spending over $60 billion, while xAI/SpaceX has reported sizable AI-related losses. Analysts caution Anthropic’s profit may be fragile given ongoing, large multi-year compute purchase commitments with providers like Google, AWS, and SpaceX.