One AI Bet Made $45 Billion This Year, and HR Teams Should Care

The biggest AI story this week did not happen in a model release. It happened on a balance sheet. A single stake in one frontier lab swung a global holding company to a record profit. That is the kind of AI portfolio valuation event that resets expectations for HR and finance teams alike. The fundraising bar just moved. So did the talent bar. Three other stories sharpen the picture this week. First, a UK regulator threatening fines on AI hiring tools. Second, a deeper Workday and Microsoft handshake inside Copilot. Finally, Anthropic publicly picking a side in the US-China AI race.

The AI Portfolio Valuation Trade That Reset a Global Profit Line

SoftBank Group reported fiscal-year results on May 13. Net profit hit ¥5 trillion (about $31.8 billion), up 333.7% year over year. Meanwhile, the Vision Fund alone booked a $46 billion gain on the year. Almost all of it came from a single position: OpenAI. (Source: CNBC)

What happened

SoftBank’s cumulative investment in OpenAI sits at $34.6 billion. The fair value of that stake on March 31 was $79.6 billion. That is a $45 billion unrealized gain in twelve months. Specifically, OpenAI’s headline valuation moved from $157 billion in October 2025 to $852 billion by March. That is a 5.4x revaluation in five months. For context, OpenAI now runs at a reported $2 billion monthly revenue. In addition, SoftBank has committed to push total OpenAI exposure to $64.6 billion through a $30 billion follow-on. The $10 billion tranches are scheduled in April, July, and October 2026. (Source: Yahoo Finance)

Why this matters for HR leaders and founders

This is the moment the AI portfolio valuation story stops being a venture-capital narrative. Instead, it becomes a public-markets narrative. As a result, every public CFO with an AI line item now has cover to spend harder. One fund’s AI exposure just produced a $46 billion swing. That trickles into your world fast. For example, your AI vendors will raise prices to chase margin. Your engineering comp benchmarks will tick up again. Frontier labs and their portfolio companies are still hiring aggressively.

There is also a softer signal here. The market is rewarding companies with concentrated, well-timed AI exposure. So founder pressure on HR teams to “show your AI bets” will keep escalating. Roadmap reviews will look at AI hiring velocity. They will look at agent deployment counts. They will look at internal automation metrics. Build the reporting now, because boards will ask for it by Q3.

What to do this week: Pull your year-over-year AI tooling spend into one slide. Add your AI-adjacent hires. Add your top three internal AI deployments. You will be asked for it. In addition, refresh your view of the AI tools your HR stack actually uses. Then you can defend or rebalance the spend with data, not vibes.

UK Regulator Tells Employers: Rubber-Stamping AI Hiring Is Not Compliance

The UK Information Commissioner’s Office warned employers on May 15. Specifically, AI-powered screening, candidate ranking, and video-interview analysis may already breach UK data protection law. The threshold: humans must be meaningfully involved in every consequential decision. (Source: TechTimes)

The ICO drew evidence from more than 30 UK employers. The finding is blunt. Most employers claim “decision support” with a human making the final call. However, in practice the human rubber-stamps the AI-generated shortlist. They lack authority, discretion, or time to override it. Therefore, per the ICO, that is not meaningful involvement. It triggers the automated decision-making rules under UK GDPR. The consultation closes May 29, 2026. (Source: ICO)

So what? If you hire into the UK, even remotely, audit your AI applicant tracking system workflow this quarter. Specifically, document where a human reviewer can actually reject the model’s recommendation. Time-box the review. Capture the rationale in writing. Vague “human in the loop” language will not survive a regulator’s audit.

Workday Sana Lands Inside Microsoft 365 Copilot, So HR Lives Where Work Happens

Workday made its Sana Self-Service Agent generally available inside Microsoft 365 Copilot on May 13. Employees can now ask Copilot how much vacation they have left. They can file the time-off request. They can check expense status. They can prepare for a review. They can pull a family-leave policy. All of it happens inside Teams or Outlook. Meanwhile, the agent securely calls back into Workday. It respects existing approvals and role-based permissions. Then it returns the result inside Copilot. (Source: Workday Newsroom)

So what? This is the second major HCM platform to plant a foothold in the productivity surface this quarter. For HR ops running both Workday and Microsoft 365, the practical question shifts. It is no longer “do we need a self-service portal?” Instead, the question is which agent owns the conversation when a manager asks two things in one chat. Start sketching the routing policy now. Moreover, if you run a leaner stack, check how your tooling exposes AI agents for HR workflows. Copilot and equivalents will keep absorbing transactional HR work.

Anthropic Picks a Side in the AI Race and Pushes for Tighter US Compute Controls

Anthropic published “2028: Two Scenarios for Global AI Leadership” on May 14. The policy paper sketches two futures. In the first, the United States tightens export controls. It also disrupts distillation attacks and accelerates allied AI adoption. In the second, it does not. Then authoritarian regimes set the norms. Specifically, the lab calls 2026 the breakaway window. It urges policymakers to crack down on chip smuggling and overseas compute access. (Source: Anthropic)

So what? The political climate around AI compute is getting louder. That has direct talent-strategy implications. For instance, if you hire AI engineers globally, expect more friction. Cross-border data, model weights, and fine-tuning data movement will all face new compliance overhead over the next 12 months. Therefore, founders building India- or APAC-based AI teams should plan for it. The AI portfolio valuation upside is also fragile here. Export-control whiplash can re-rate frontier labs in either direction.

Quick Hits

• Salesforce will spend around $300 million on Anthropic tokens this year, almost entirely on coding workloads. CEO Marc Benioff said so on the All-In podcast. Moreover, he flagged the need for an intermediary routing layer between cheap and frontier models. (Source: The Next Web)
• Workday CTO Peter Bailis stepped down to join Anthropic as a member of technical staff. He will work on reinforcement learning. As a result, a senior HCM-AI architect now sits inside a frontier lab that is openly exploring HR software. (Source: The Information)
• The EU AI Act Omnibus deal was politically agreed on May 7. It pushes the stand-alone high-risk AI compliance deadline to December 2, 2027. In addition, SME relief extends to companies with up to 750 employees and €150 million in revenue. (Source: EU Council)

The Bottom Line on This Week’s AI Portfolio Valuation Story

A $45 billion paper gain on one AI stake is the kind of AI portfolio valuation signal that pulls budgets and roadmaps the same way. As a result, your AI vendors will get more expensive. Your hiring tools will get more scrutinized. Your CFO will ask harder questions about AI ROI. So if you are building a globally distributed team, look for HR tooling that keeps pace with this curve. That means agent-friendly APIs and multi-country payroll. Asanify is worth a look.

FAQ

What is an AI portfolio valuation gain and why does it matter for HR teams?

An AI portfolio valuation gain is the unrealized increase in fair value of equity held in AI companies. It matters for HR teams because outsized gains keep pushing budgets up. For example, SoftBank booked a $45 billion mark-up on its OpenAI stake in the fiscal year ended March 2026. As a result, that pressure flows directly into compensation benchmarks, AI vendor pricing, and board expectations on AI deployment.

Are AI hiring tools legal in the UK in 2026?

AI hiring tools are legal in the UK. However, the Information Commissioner’s Office warned on May 15, 2026 that many deployments may breach UK GDPR. Specifically, the human reviewer must be able to meaningfully change the outcome. Approving an AI shortlist without time or authority to reject it counts as rubber-stamping. Therefore it triggers automated decision-making rules. The ICO consultation closes May 29, 2026.

What changed in the EU AI Act Omnibus deal of May 2026?

On May 7, 2026, the EU Council and Parliament reached a political agreement to extend high-risk AI compliance deadlines. Stand-alone high-risk systems now have until December 2, 2027. AI embedded in regulated products has until August 2, 2028. In addition, the simplified compliance framework now covers companies with up to 750 employees and €150 million in annual revenue. Formal adoption is expected before August 2, 2026.

Not to be considered as tax, legal, financial or HR advice. Regulations change over time so please consult a lawyer, accountant  or Labour Law  expert for specific guidance.