Record spend, record caution
Enterprise spending on generative AI more than tripled in a single year, from $11.5 billion in 2024 to $37 billion in 2025 (Menlo Ventures). Over the same period, MIT's Project NANDA found that 95% of the organisations it studied saw no measurable return from generative AI, and only 5% of custom enterprise AI tools reached production (Virtualization Review on MIT NANDA).
That gap is the most important fact for any specialist firm selling into the enterprise. The money is there. The confidence is not, and large organisations respond to that the way they always respond to risk: by becoming far more selective about who they trust.
The buyer is asking a different question
It is a group decision. Complex B2B purchases now involve 6 to 10 decision-makers, and 77% of buyers describe their latest purchase as very complex or difficult (Gartner, via The Starr Conspiracy). Every one of those people can say no.
It is slow. 6sense's 2025 study of nearly 4,000 buyers puts the average purchase at more than 10 people and 10.1 months (Sales & Marketing Management). In large enterprises, MIT found AI pilots take nine months or longer to reach full implementation, against about 90 days in mid-market firms (RCP Magazine).
It is driven by fear. In the same 6sense study, 70% of buyers said economic anxiety directly influenced their vendor choice, often in favour of incumbents (6sense via Business Wire).
In cyber, the risk is now personal. Under NIS2, members of the management body must approve and oversee cybersecurity measures and can be held personally liable. Regulators can temporarily ban executives of essential entities from management roles (Bastion). Under DORA, the management body of a financial entity carries ultimate responsibility for ICT risk (Compliance Hub).
Put together, the enterprise buyer has stopped asking "who is best?" It is asking "who is safest?" And safest usually means someone they already know.
Why excellent specialist firms still lose
The deal is decided before anyone calls you. Buyers spend only 17% of their purchase journey meeting suppliers (Gartner, via The Starr Conspiracy). 6sense found that 94% of buying groups rank their preferred vendors before first contact, and fill four of their five shortlist places on day one (CustomerThink).
The favourite almost always wins. The vendor a buyer prefers before engaging sellers wins 80% of deals (6sense via Business Wire). Do the arithmetic: if the favourite takes 80%, the other four shortlisted firms share the remaining 20%, roughly 5% each.
Relationships build the shortlist. 85% of buyers had prior experience with the vendor they chose, and 75% personally knew sellers at the vendors they evaluated (Sales & Marketing Management).
That is the trap for a specialist firm of 40 to 150 people. Answering a cold RFP usually means joining a shortlist built around someone else. Being technically better rarely changes the outcome, because the decision was shaped in rooms you were not in.
When buyers cannot tell specialists apart and fear getting it wrong, the default winner is the large systems integrator. Scale feels safe, even when the senior people in the pitch are not the people who deliver.
What a consortium changes
The enterprise has already decided to buy from specialists. 76% of AI use cases were purchased rather than built in 2025, up from 53% a year earlier (Menlo Ventures). The evidence backs that choice: MIT found projects bought from specialised partners succeeded about 67% of the time, roughly twice the rate of internal builds (Xpert Digital on MIT NANDA).
So the question is no longer whether enterprises buy from specialists. It is which specialists they trust. A governed consortium answers each of the buyer's fears directly.
| The buyer's fear | How the consortium answers it |
|---|---|
| "We don't know these firms." | Opportunities start in Prescient Circle's personal advisory boards and roundtable dinners with enterprise CIOs, CTOs and CISOs, long before a formal RFP. |
| "Who carries the risk?" | One qualified lead partner is accountable for each programme, and every other partner stands behind its own scope. |
| "Can a small firm deliver at our scale?" | Three to five specialists are combined for each engagement, with pooled case studies as proof. |
| "Is this compliant where we operate?" | Regulatory partners sit in the US, UK and EU, where the contract is signed. |
| "Will it actually be adopted?" | Change and adoption is a core discipline in every team, because that is where most AI programmes fail. |
| "Who is really in the room?" | Named senior specialists, chosen for the engagement, deliver the work they pitched. |
What it means for a member firm
For a specialist firm, the difference is where it enters the buying journey, and with whom.
| Selling alone | Inside the consortium | |
|---|---|---|
| Where you enter | After first contact, which 6sense puts at 61% of the way through the buyer's journey (CustomerThink) | In the room with enterprise technology leaders before an RFP exists |
| Your odds | One of five names on a shortlist whose favourite wins 80% of the time | Part of the team the buyer already knows |
| Deal size | What one firm can carry on its own | Multi-disciplinary programmes, typically $5 million to $15 million |
| Buyer trust | Your own track record | One accountable lead, pooled case studies and local regulatory partners |
| Cost of a bid | Carried by you alone | Shared across the three to five partners on that bid |
| Who you work with | Whoever else answered the RFP | Vetted firms on a register every member can see |
One programme changes a firm of this size. A $7 million engagement shared across five partners is about $1.4 million of work each, won from a relationship rather than a cold bid.
The window is now
Enterprise AI spending has tripled in a year, while buyers narrow the circle of firms they trust. The specialists inside that circle when budgets move will win the next wave of programmes. Those outside it will keep answering RFPs that were decided before they arrived.
Prescient Circle is building two consortiums, one for AI and one for Cyber, around exactly this shift. We are vetting founding members now: firms based in the US or Europe, with real depth in one discipline and references that will stand up to an enterprise buyer's due diligence.
Founding applications close on Nov 9, 2026.