DX Today AI Daily Brief - Friday, May 8, 2026

DX Today AI Daily Brief - Friday, May 8, 2026

NVIDIA invests up to $2.1 billion in data center operator IREN to deploy 5 gigawatts of AI infrastructure. The European Parliament and Council reach political agreement on a Digital Omnibus delaying the toughest parts of the EU AI Act to December 2027. Anthropic taps SpaceX for 300+ megawatts of compute via the Colossus 1 facility. Atlassian opens its Teamwork Graph at Team 26 with 150 billion connections. Moonshot AI raises $2 billion at a $20 billion valuation. Google partners with Elementl Power on 1.8 GW of advanced nuclear. OpenAI launches three voice models translating 70 languages live. Zyphra ships an 8B reasoning model trained on AMD hardware. NANO Nuclear and Supermicro sign a microreactor MOU. AI insurance startup Corgi raises $160M Series B. Elucid wins the cardiology MedTech Breakthrough award for Plaque-IQ. And Salesforce publishes new research on enterprise agentic harness reliability.

DX Today - Your daily AI news briefing, delivered every morning.

Send us fan mail

#AINews #ArtificialIntelligence #AIInfrastructure #EUAIAct #DXToday

Transcript

SPEAKER_03 0:00

It's Friday, May 8, 2026. You're listening to the DX Today AI Daily Brief. Today, Nvidia pours up to$2.1 billion into data center operator IREN. Brussels strikes a political deal to delay the toughest parts of the EU AI Act by more than a year. And Anthropic taps Elon Musk's SpaceX for 300 megawatts of compute. Let's get into it.

SPEAKER_00 0:28

Nvidia and Australian listed data center operator IRE Annon announced a sweeping new partnership Thursday. Under the agreement, NVIDIA will deploy up to five gigawatts of its DSX-branded AI infrastructure across Irene and facilities, with the Sweetwater Texas campus as the anchor site. As part of the deal, IRN granted NVIDIA a five-year right to acquire up to 30 million ordinary shares at$70 apiece, an option that, if fully exercised, would inject roughly$2.1 billion into the firm. Investors reacted fast. IRN shares spiked more than 27% intraday before settling. The deal cements IRAN's role as a leading neo cloud, renting NVIDIA-powered capacity to hyperscalers hungry for compute.

SPEAKER_03 1:16

Across the Atlantic, a regulatory reshuffle.

SPEAKER_05 1:20

The European Parliament and the Council of the European Union reached a provisional political agreement Thursday on the so-called digital omnibus on AI, a sweeping amendment to the landmark EU AI Act. The compromise pushes back compliance deadlines for high-risk AI systems listed in Annex III, including biometrics, critical infrastructure, and employment tools. Those obligations, originally due August 2nd, of 2026, now take effect December 2nd, 2027, for standalone systems. Lawmakers also added a new prohibition on AI-generated non-consensual intimate imagery and child sexual abuse material. The deal extends regulatory exemptions for small mid-cap firms, postpones national AI sandbox deadlines, and tightens transparency rules for synthetic content from six months to three.

SPEAKER_02 2:23

Now to a startling compute deal. Anthropic announced a major compute partnership with Elon Musk's SpaceX, gaining access to the full Colossus 1 AI data center in Memphis, Tennessee. The agreement gives Anthropic over 300 megawatts of capacity, anchored by more than 220,000 NVIDIA GPUs, including H-100, H-200, and the next generation GB200 accelerators. Anthropic says the new compute will power its Cloud Pro and Cloud Max assistants, lifting the five-hour rate caps that frustrated paid subscribers. The deal also opens a longer-term track for space-based data centers, a long-stated Musk ambition. Notably, the partnership lands amid Musk's ongoing lawsuit against OpenAI, signaling a striking realignment in AI infrastructure alliances.

SPEAKER_03 3:18

From Anaheim, an enterprise AI opening.

SPEAKER_01 3:22

Atlassian threw open its teamwork graph at the Team 26 conference, a move designed to let AI agents from any vendor reason and act across enterprise tools. The graph now contains more than 150 billion connections linking people, projects, documents, and decisions across Atlassian and third-party platforms. Atlassian is releasing two new interfaces in Open Beta, a teamwork graph command line interface, and a model context protocol server delivered through its Rovo AI assistant. The company also announced general availability of agents in Jira, expanded Jira product discovery, and a remix with Rovo feature for Confluence that converts pages into charts, timelines, and full presentation decks on demand.

SPEAKER_03 4:10

Now to Beijing, a mega round.

SPEAKER_04 4:12

Chinese AI startup Moonshot AI, the developer behind the Kimi chatbot, closed a roughly$2 billion strategic funding round Thursday, lifting its valuation past$20 billion. Maytoon's Long Z investments led the deal, joined by Watere Capital Management, China Mobile, and CPE Funds Management. Moonshot says annual recurring revenue surged from over$100 million in early March to more than$200 million in April, on the back of its updated Kimi K 2.5 model. The round is Moonshot's third in 2026, bringing total fundraising this year past$3.9 billion, as Beijing's open source AI champions navigate tighter domestic IPO rules.

SPEAKER_00 5:01

Powering AI with atoms. Google announced an agreement with Elemental Power to pre-position three advanced nuclear sites across the United States, each designed to deliver at least 600 megawatts. Combined, the trio could yield 1.8 gigawatts of clean baseload electricity dedicated to feeding Google's voracious AI data center expansion. Google will fund early phase development, including site permitting, regulatory engagement, and commercial structuring, while securing an option to purchase the eventual output. Elemental, founded in 2022, aims to deploy more than 10 gigawatts of advanced nuclear capacity by 2035. The deal lands as hyperscalers scramble to lock in around-the-clock carbon-free power for an AI workload pipeline straining the grid.

SPEAKER_05 5:50

Now to OpenAI. OpenAI rolled out three new voice models capable of live translation across 70 languages in a release positioned squarely at enterprise contact centers. Real estate giant Zillow ran early pilots and reported a 26 percentage point jump in successful call outcomes once the models were deployed. Alongside the voice rollout, OpenAI widened its Chat GPT advertising pilot, launching a self-serve ads manager that allows small businesses across the United States to set budgets, upload creative assets, and launch campaigns directly inside the OpenAI portal without engaging a partner agency. Together, the launches signal OpenAI's deepening push to monetize ChatGPT through both real-time voice infrastructure and a native advertising stack.

SPEAKER_02 6:50

An open weight challenger arrives. Open source lab Zifra released Zia 1-8B, a reasoning-focused model trained entirely on AMD instinct hardware. Zifra says the 8 billion parameter model outperforms several open weight rivals, many times its size on math and structured reasoning benchmarks. The release is a milestone for AMD's data center GPU strategy, demonstrating that high-quality frontier class training runs no longer require NVIDIA Silicon. Zifra is positioning ZIA as a permissively licensed alternative for enterprises building agentic AI systems where reasoning reliability, not raw scale, is the bottleneck. The launch arrives as companies aggressively diversify training infrastructure to reduce single-vendor exposure and lower the unit economics of large-scale AI deployment. Microreactors for the Cloud.

SPEAKER_01 7:47

Nano Nuclear Energy and Supermicro Computer signed a strategic memorandum of understanding to integrate Nano's advanced micro reactor systems with Supermicro's high-density AI server racks. The framework covers joint go-to-market activities, mutual sales referrals, and co-engineered deployments anchored by Nano's Kronos Modular Reactor Design. The companies plan combined offerings spanning micro reactors, server infrastructure, cooling, and lifecycle services aimed at hyperscale and edge AI operators struggling to secure firm power. The partnership reflects a broader pivot in the sector. With Grid Interconnect queues stretching years, AI infrastructure providers are increasingly betting that small, dispatchable nuclear assets co-located with data centers will become a credible answer to runaway compute demand.

SPEAKER_03 8:40

Insurance gets agentic.

SPEAKER_04 8:42

AI insurance startup Corgi closed a$160 million Series B funding round, lifting its post-money valuation to$1.3 billion. Corgi runs an AI agent platform that streamlines small business insurance, payroll, and trucking workflows, and plans to use the new capital to expand aggressively into all three verticals. The deal lands amid a broader investor surge into vertical AI agents that automate paperwork-heavy back office tasks long resistant to digitization. Corgi's leadership says the funding will support product expansion, deeper carrier integrations, and significant hiring of risk and underwriting talent across United States markets, where premium volume keeps climbing on the strength of small business formation.

SPEAKER_03 9:32

A cardiology breakthrough.

SPEAKER_00 10:12

Together, the wins underscore AI's rapid embedding across the clinical workflow, from imaging and diagnostics to documentation and remote patient monitoring.

SPEAKER_03 10:23

Salesforce on agent reliability.

SPEAKER_05 10:26

Salesforce published new research Thursday, arguing that not all agentic harnesses, the orchestration layer that lets an AI agent act autonomously, are created equal. Salesforce contends an enterprise grade agent must understand not just an individual user's intent, but the broader organizational context, shared data history, and policy guardrails that determine whether the agent has standing to act. The piece zeroes in on multi-step task completion, citing reliability gaps in current frameworks. The release lands amid intensifying competition over enterprise AI agent platforms, with Microsoft ServiceNow and Atlassian each pitching governance first architectures as the path to deploying autonomous agents safely at scale across regulated industries.