Friday, 2 October 2026European Markets

Everyone Plans to Deploy AI Agents. The Harder Question Is Whether the Data Is Ready

A survey reported by MIT Technology Review says 100% of respondents plan to use agentic AI within two years, yet the supporting data is weaker than the headline. The source has a poor measured reliability record. Nvidia's audited accounts and the Manulife, Box and Nomura announcements are firmer ground.

LM Salvado
LM Salvado

October 2, 2026

Everyone Plans to Deploy AI Agents. The Harder Question Is Whether the Data Is Ready
Image generated by AI for illustrative purposes. Not actual footage or photography from the reported events.
Loading stream...

A note on scope for European readers: this dossier contains no EU companies, no EU regulation and no European market data. Every company in it is North American, Japanese or US-based, so we have not stretched it into a European story. What follows is the global picture that European businesses are buying into, with the governance questions that matter most to them.

The headline: universal intent, shaky evidence

The most quoted claim in this story is that, within two years, 100% of surveyed organisations plan to use agentic AI, with 69% expecting to use it widely.1 The same piece says that in organisations classed as 'data laggards', AI access to company data falls to 30% or less.1 Via News's own summary of the coverage also cites figures of about 45% of company data being accessible to AI and only around half of organisations trusting their agents' decisions, attributed to a Google Cloud survey. Those two figures are not among the claims we could check against the source text. Treat them as unverified.

The source needs the same caution. The article is "Scaling AI agents with trustworthy data" from MIT Technology Review. Our measured reliability record for it is poor: 0% of 11 checked claims held up.1 That is a small sample, and it does not mean the survey is wrong. It does mean you should not treat any single percentage from it as settled.

What the audited numbers can and cannot tell you

The only fidelity-checked numbers in our dossier come from SEC filings for Nvidia, the chipmaker whose tools sit under much of this agent ecosystem. The dossier links Nvidia to products such as NeMo Guardrails, the NeMo Agent Toolkit and the Nemotron 3 Super model, and to customers including Mount Sinai Health System and Yum! Brands.

Nvidia's cost of revenue, the direct cost of what it sells, rose from $16.621 billion in fiscal 2024 to $32.639 billion in fiscal 2025 and $62.475 billion in fiscal 2026.2 That is more than a threefold rise in two years. It is also about six times the company's fiscal 2026 cash balance of $10.605 billion.3 We have no outside comparison in the dossier to make these sums feel larger or smaller, and we will not invent one.

Cash is less steady. Nvidia held $7.28 billion in fiscal 2024 and $8.589 billion in fiscal 2025.3 Quarterly figures swing more: $15.234 billion at the first quarter of fiscal 2026, falling to $11.639 billion in the second and $11.486 billion in the third, then $13.237 billion in the first quarter of fiscal 2027.3 Earnings per share were $11.93 in fiscal 2024, $2.94 in fiscal 2025 and $4.90 in fiscal 2026.4 The dossier does not explain that swing, so do not read it as a collapse in profit. We cannot say whether you already own Nvidia through a fund, because the dossier carries no index weightings.

None of this proves enterprise AI demand. It shows the supplier side is spending heavily to serve it. Whether customers can use what they buy is the question the rest of the dossier addresses.

Incumbents are buying governance, not just models

The large-company announcements are about control as much as capability. Manulife and Microsoft announced a renewed five-year agreement, with Manulife adopting Microsoft's Frontier Suite, deploying Microsoft Agent 365 and expanding Microsoft 365 Copilot to more than 30,000 employees.5 Manulife's Shamus Weiland said the partnership is "a critical enabler of Manulife's continued evolution into a truly AI-driven organization".5 The release was headlined as accelerating enterprise AI governance as well as innovation.5 Its source, NewsEOD, has a measured reliability of 57% of 4,954 checked claims holding up.5 That is a press-release feed, so read it as what the companies say about themselves.

Box announced agent guardrails, third-party agent activity oversight, prompt injection detection and agent classification-based access policies.6 Nomura Research Institute's Tatsutoshi Murata said he expects Box to "provide the administrative features needed to safely leverage this new era of AI".6 The NewsEOD feed's 57% reliability applies here too.6

Our analysis: the pattern across both announcements is that large organisations are spending on controlling agents, not only on building them. That fits the trust gap the survey describes, though the survey's own numbers remain unverified.

Startups are pricing against labour, not software

Several vertical startups define their markets by the work humans do today. Penguin AI's Glenn Herzberg says US healthcare administration runs about a trillion dollars a year, about a quarter of total health spend, with published estimates putting around $570 billion of it in work that has no effect on health outcomes.7 That $570 billion is a published estimate quoted by a company with a commercial interest in the number.

Covecta's Ben Thomas says its addressable market is tens of thousands of financial institutions globally, and that "we are not just disrupting their software budget but their labor budget as well".8 Covecta currently serves customers across the US and UK and aspires to go global.8 The UK is the only European market in this dossier. It is also outside the EU.

Maisa AI's David Villalon describes his market as process automation of core business tasks at regulated industries, covering back office, operations and finance, where work must be auditable, reproducible and hallucination resistant.9 That requirement lines up with what the governance vendors above are selling.

Investors are paying attention to the same pattern. Primary partner Emily Man said of the Casap founders that they had "experienced the pain points of disputes firsthand at their respective large fintech companies" and saw how much internal effort it took to solve them even with a strong engineering team.10 That is one investor describing one portfolio company, not market-wide evidence.

What to watch

  • Whether the 45% and 'around half' figures can be traced to a primary survey document. Until they can, lean on the checked claims.
  • Nvidia's quarterly filings. Cost of revenue and cash are the checked indicators of how hard the supply side is being pushed.
  • Governance product launches. Box, Microsoft and Nvidia all now sell agent controls, which suggests customers are asking for them.
  • European coverage. The dossier has no EU regulatory material. When it does, that is where this story becomes directly relevant to European businesses.

Source documents

Via News is a conduit. We point to the source documents behind this report — we don't replace them. Trace any claim to its source and decide what to trust. How we source

Source Trace Score12 source documents12 with a live linkVerifiability: Strong
  1. [1]News articleYahoo Finance· July 21, 2026
    Box Unveils New Controls to Secure AI Agents Operating Across Enterprise Content
  2. [2]News articleCB Insights
    CEO Interview: David Villalon, Maisa
  3. [3]News articleCB Insights
    Executive Interview: Covecta
  4. [4]News articleCB Insights
    Executive Interview: Penguin AI
  5. [5]News articleCB Insights
    Investor Interview: Primary on Casap
  6. [6]News articleYahoo Finance· July 22, 2026
    Manulife Expands Partnership with Microsoft to Accelerate Enterprise AI Governance and Innovation
  7. [7]News articleMIT Technology Review
    Scaling AI agents with trustworthy data
  8. [8]News articleMotley Fool
    C3.ai CEO Thomas Siebel Sells 453,000 Shares for $4.8 Million
  9. [9]News articleCB Insights
    CEO Interview: Arlo
  10. [10]News articleCB Insights
    CEO Interview: Binary World
  11. [11]News articleCB Insights
    CEO Interview: LARX
  12. [12]News articleCB Insights
    CEO Interview: Shepherd
LM Salvado
LM Salvado

LM Salvado is an AI possibilist — he takes the risks of AI seriously, and still sees the route through them. Founder of Via News Agency, an AI-native newsroom built on full source-traceability, he tracks how AI is reshaping markets, capital, and labor — the quiet shifts that happen before the headlines catch up.

What we know · the intelligence behind this page
Live from the substrate
What we're seeing
AI-Driven Drug Development Meets Biotech Deal-Making and Regulatory Catalysts
AI-designed therapeutics (Insilico's rentosertib showing biological-age reductions) are moving into the clinical mainstream. Large-cap biotech is simultaneously reallocating capital through M&A (Lilly–Merida, $2.9B) and government funding (BARDA–Basilea), while trial failures (ziltivekimab, a 9.4% Novo Nordisk share drop) and upcoming FDA catalysts (the ivonescimab PDUFA on 2026-11-14) drive volatility. The wider AI regulatory and legal climate (Tesla Cybercab probe, xAI's Minnesota loss, OpenAI suits) is tightening, though QAIAx's micro-cap trial claims are speculative and weakly connected.
Our read on the data ›
Signals we're tracking
Satellite-Terrestrial Network Integration Acceleration
Increased investment and launches in hybrid satellite-cellular networks across telecom industry; competitive responses from other carriers; regulatory activity around satellite spectrum; expansion of emergency/rural connectivity use cases
Patterns we're watching ›
Where sources disagree
ING Group
Both facts record the same metric (shares_outstanding) for ING Group at the identical observation date (2025-12-31). FACT A states 2,902,437,688 shares; FACT B states 2,902 million shares (2,902,000,000). The difference is 437,688 shares (~0.015%). This is a genuine value conflict, though the discrepancy appears to result from FACT B rounding to the nearest million while FACT A provides the precise count.
We flag conflicts openly ›
Recently verified
✓ Checked against the original source
4,985
facts traced to their source — and we flag the ones that don't hold up.
101 entities tracked4,985 facts checked against source5,327 source documents archived
Query this data → isubstrate.com