The Brief · Issue 7

Week of 10 August 2026 — Databricks' $5B round, and the gap between AI spend and AI readiness

Databricks raised $5B at a $190B valuation, Alteryx found half of IT leaders can't get business context into AI, and a new Dutch cyber law took effect.

September 22, 2026

Compiled on 22 September 2026 for The Brief's launch archive, from sources published between 10 and 16 August 2026.

Four stories that mattered

Databricks raises $5 billion at a $190 billion valuation

Databricks closed a $5 billion strategic funding round on 13 August, led by Coatue with participation from Blackstone, MGX, T. Rowe Price and more than a dozen other investors, valuing the company at $190 billion. Databricks said its revenue run-rate has passed $7 billion, growing more than 80% year over year. CEO Ali Ghodsi said enterprises "want agents working across their business that remember context, deliver accurate answers, and execute work without blowing through budgets."

The round is one of the largest private funding events of the year for a data company and a strong signal of continued investor confidence in the data-platform-plus-AI-agent bet, even as public scrutiny of AI ROI grows (see below).

Source: Databricks, "Databricks Grows >80% YoY, Surpasses $7B Revenue Run-Rate, Scales Lakebase, Genie, and Unity AI Gateway"

Alteryx: half of IT leaders can't get business context into their AI systems

Alteryx published its 2026 IT Leader Research on 13 August, a survey of 1,400 IT and automation leaders conducted by Coleman Parkes. While 77% agreed that business context — the rules, definitions and operational knowledge that shape how a company actually works — is critical to accurate AI output, 53% said they struggle to bring that context into the systems and workflows their AI relies on. 80% expect AI spending to increase over the next two years; only 69% report moderate or significant ROI so far.

This is a concrete, survey-based version of a complaint analytics teams have made for years about self-service BI: giving a system access to data isn't the same as giving it a shared understanding of what that data means.

Source: Alteryx via PR Newswire, "53% of Organizations Struggle to Translate Business Context Into AI Despite Rising AI Investment"

Research: ChatGPT often decides which brand to recommend before it searches

A study published in Search Engine Journal on 14 August found that ChatGPT frequently writes specific brand names into its own web-search queries before fetching any pages, based on an analysis of dozens of real conversations captured via browser developer tools. Brands that appeared in ChatGPT's self-generated query were mentioned in its final answer 68.9% of the time; brands whose pages were fetched but never named in the query were mentioned only 2.1% of the time — roughly a 33-times difference. In 21 of 27 first-query tests, ChatGPT's initial search already contained brand names the user had never typed.

For teams measuring "share of voice" in AI answers, this suggests the decisive moment often happens before retrieval at all, in which brands the model already associates with a category — a training-data effect that on-page optimization can't fix after the fact.

Source: Search Engine Journal, "ChatGPT Already Knows Who It'll Recommend Before It Searches"

The Dutch Cybersecurity Act takes effect, with no transition period

The Netherlands' Cyberbeveiligingswet (Cybersecurity Act), implementing the EU's NIS2 directive, entered into force on 15 August with no general grace period, immediately applying registration, duty-of-care, incident-reporting and board-level governance obligations to an estimated 8,000-plus organizations. Higher education institutions are the one exception, with a three-year transition window.

Any analytics or data team operating essential or important infrastructure in the Netherlands is now on the clock for incident reporting and governance duties that touch the same data pipelines and access controls analytics relies on.

Source: Houthoff, "Dutch Cybersecurity Act enters into force 15 August"

Tool moves

One how-to

Give AI a shared glossary before giving it more access. The Alteryx finding above — 77% say business context matters, 53% can't operationalize it — points to a specific, fixable gap: most organizations have tribal knowledge about what "active customer" or "qualified lead" means, but nowhere an AI system (or a new analyst) can look it up. A basic build-out:

  1. Start with the terms that already disagree. Pick the five metrics different teams define differently — "active user," "churn," "revenue" — before trying to document everything.
  2. Write one definition per term, with the calculation. A business glossary entry needs the plain-English meaning and the exact formula or filter that produces it, not just a description.
  3. Attach it to the semantic layer, not a wiki page. A definition a person has to go find is a definition an AI copilot will never see; it needs to live next to the data it describes, the way Databricks' new Pages feature or a data dictionary does.
  4. Version it like code. Definitions change — who counts as an "active" user shifts with the product — so track changes and dates, the same discipline used for data lineage.
  5. Make it the first thing a new AI tool reads. Before connecting a copilot or agent to a data source, point it at the glossary first; an agent that can quote your definition of a metric back to you is far less likely to misuse it.

This doesn't require a platform purchase — a well-maintained spreadsheet beats an undocumented one — but it does require someone to own it.

One number

3.4% — the year-over-year increase in the U.S. Consumer Price Index for July 2026, reported by the Bureau of Labor Statistics on 12 August. Shelter rose 3.2% annually and accounted for roughly two-thirds of the month's 0.1% increase; energy was up 14.7% year over year, driven largely by a 24.6% jump in gasoline prices; core inflation, excluding food and energy, rose 2.5% annually.

The headline figure is a national average: it says nothing about price changes in any one city, income bracket or spending category, and a single month's reading can be revised or reversed by the next release.

Get the next one