The Conference Business Didn't Shrink. It Changed Hands.

The story a lot of people in technology tell each other is that the conference business is in trouble. Budgets cut, wars disrupting travel, events dying quietly. It is an easy story because the visible signals fit it: through 2025 and 2026, conferences moved city, changed month, changed owner and shrank their side-event weeks, often several at once.

The story does not survive the numbers.

The claim that fails on contact

Below are the flagship technology events for which comparable multi-year attendance exists. Five of six grew or held. The one clear decline still sat above its 2024 level and did not lose a single exhibitor.

Event	2024	2025	2026	Direction	Figure type
CES	138,789	142,465	148,392	Up three years	Independently audited
VivaTech	165,000	180,000	200,000+	Up three years	Organiser
gamescom	335,000	357,000	—	Up	Organiser and trade body
IFA Berlin	—	~220,000	—	Exhibitors up 5%	Organiser
Web Summit	71,528	71,386	—	Flat	Organiser
MWC Barcelona	101,000	109,000	~105,000	Down 4%	Organiser

CES is the one to weigh most heavily, because the Consumer Technology Association commissions an independent attendance audit rather than publishing a marketing number. It has grown three years running. A dash means the 2026 edition had not happened when this was written; none of those cells contains a projection.

The financial side agrees. Informa's B2B live events division grew revenue 9.5 per cent on an underlying basis in 2025 and around 8 per cent in the first half of 2026. RX, the exhibitions arm of RELX, grew 8 per cent underlying in 2025 on revenue of £1.186bn and told investors it is positioned for further growth in face-to-face events. Those are audited filings from listed companies, not press releases.

An industry in structural decline does not look like this.

Force one: the industry is being bought

The clearest structural change is ownership. Events that spent a decade being run by their founders are being absorbed into large exhibition groups, and the buyers are paying real money.

Informa acquired Ascential, owner of Money20/20 and Cannes Lions, for approximately £1.2bn, completing in October 2024. In July 2026 Hyve Group acquired three events at once: RAISE Summit, Machina Summit and Signal Week, the event that ran for eight years as Paris Blockchain Week. No price was disclosed for the Hyve deals.

The instinct is to read acquisition as distress. The evidence does not support it. Informa described Money20/20 as a growth opportunity and later called its first Middle East edition in Riyadh one of its largest-ever launches. Hyve framed its purchases as building an AI "super vertical" and strengthening a fintech portfolio. Signal Week itself reports more than 10,000 attendees from over 100 countries, with 70 per cent of the room at C-suite level. Nobody buys that because it is failing.

What consolidation does change is who the event is designed for. A founder-run conference optimises for the community it came from. A portfolio-owned conference optimises for what the portfolio is good at, which leads directly to the second force.

Force two: the product stopped being attendance

The most consistent signal across every event we examined is that organisers have stopped selling a crowd and started selling a schedule.

Hyve's stated reason for buying Signal Week was to bring across the curated meeting engine behind Fintech Meetup, which runs more than 50,000 double opt-in meetings. Signal Week already reports over 53,000 meetings around its own audience. Slush expects more than 25,000 meetings across two days and advertises over 1,000 pre-scheduled meetings an hour, from a capped audience of 12,000 to 13,000. TOKEN2049 runs matchmaking as core infrastructure rather than an app you might download.

Independent research points the same way. Explori, surveying exhibitors across more than 3,000 B2B events, found exhibitor satisfaction scores 27 points above 2019 levels, with exhibitors explicitly prioritising meaningful interactions and business outcomes over footfall.

Read alongside the cost data, this reframes what looks like contraction. Explori also found that 82 per cent of exhibitors planning to attend fewer shows cited exhibit-related costs, and 67 per cent said those costs were rising faster than their other costs. A company exhibiting at four shows instead of six has not lost faith in events. It has decided that two of them were not producing meetings.

The counter-example that matters

It is tempting to conclude that unofficial side-event weeks are eating the paid main events. The evidence does not support it. TOKEN2049 Singapore reported more than 25,000 attendees alongside over 1,000 side events in 2025; Dubai reported 15,000 with 500-plus. Devconnect Buenos Aires drew 14,000 people from more than 130 countries with 500-plus community events across the city. The side-event economy and the main event grew together, not at each other's expense.

Force three: geopolitics, real and local

Conflict did disrupt events in 2026. It disrupted them where the conflict was, and the effect did not generalise.

The clearest case is TOKEN2049 Dubai, postponed from 2026 to April 2027, with the organisers citing ongoing regional uncertainty and its effect on safety, international travel and logistics. Informa reported moving more than fifteen major brands out of the first half of 2026 in India, the Middle East and Africa for conflict-related reasons, and still delivered around 8 per cent underlying growth in the division.

MWC Barcelona is the useful control. The GSMA described its 2026 edition as taking place despite global travel disruption, and it still drew nearly 105,000 people, above 2024, with exhibitor numbers unchanged. Disruption showed up as a four per cent dip, not a collapse.

The visa question

Access is a genuine and unevenly distributed problem, and it is also routinely overstated. European Commission data for 2025 shows the overall Schengen refusal rate improved slightly, from 14.8 to 14.6 per cent, across more than 12 million applications. Underneath that average sit refusal rates of 40.1 per cent for the Democratic Republic of the Congo, 51.9 per cent for Senegal and 53.4 per cent for Burundi.

Both things are true: the system is not deteriorating in aggregate, and it is close to a coin flip for founders from several African countries. What does not exist, as far as we could establish, is any dataset connecting those refusals to conference attendance. The claim that visas are hollowing out international events is plausible and, at present, unevidenced.

Five things people say that the evidence does not support

"The economy broke the conference business." Listed exhibition groups reported underlying revenue growth through 2025 and into 2026, and several flagship events set attendance records in the same window. Budgets tightened; the business did not break.

"People have stopped wanting to attend giant events." CES grew three years running to 148,392 audited attendees. VivaTech went from 165,000 to over 200,000 in two years. gamescom sold out its exhibition floor for the first time in eighteen editions.

"Crypto conferences are dying." TOKEN2049 Singapore sold out above 25,000 in 2025 and Dubai grew 50 per cent year on year to 15,000. The Bitcoin Conference's own SEC filing records flagship attendance rising from 12,382 in 2024 to 31,584 in 2025. Specific events contracted, notably Consensus and ETHDenver's side programme, but that is not a sector.

"GITEX moved because of the Iran conflict." No primary source supports this. The Dubai Media Office announcement of 8 October 2025 and GITEX's own materials give commercial reasons: December tourism season, a citywide "TechCation" format, and consolidating with Expand North Star at Expo City. The geopolitical explanation may still be part of the story; it is not currently evidenced.

"Side events are killing the main conference." ETHDenver's registered side events did fall from 668 in 2025 to roughly 215 in 2026, around 68 per cent, less than the 80 per cent a January 2026 forecast predicted and never confirmed by the organiser. But TOKEN2049 and Devconnect both grew their main events while running enormous side programmes. The ETHDenver contraction was reported as budgets, a calendar clash with Lunar New Year, and visa problems, not cannibalisation.

What this means if you are choosing where to go in 2027

The practical consequence of a market that is consolidating and selling meetings rather than footfall is that the old heuristics stop working.

Size has stopped being the signal. A capped 12,000-person event with a quarter of the room in investors is a different product from a 150,000-person floor, and the second is not more valuable because it is bigger. Judge an event by who is in the room and whether you can book them.

Judge the trip on your diary, not the agenda. Where organisers have built real matchmaking, the meetings are the product and the stages are marketing. Where they have not, the value is in the side week, and you should be working it before you fly.

An ownership change is a reason to re-read the room. A portfolio buyer will optimise the event toward what its portfolio does well. That is often good, and it is rarely neutral.

Regional risk is now a booking decision. TOKEN2049 Dubai moved by a year. Check refund terms and your own government's travel advice before committing non-refundable spend to a regional event.

Treat unconfirmed dates as unconfirmed. Aggregators publish confident dates for events whose organisers have not announced them. ETHDenver 2027 is the current example.

Method, and what we could not establish

Figures here are labelled by type. Independently audited numbers and regulatory filings are treated as stronger evidence than organiser marketing, and organiser marketing as stronger than press estimates. Where an organiser publishes two different numbers, as Slush does for its own audience size, both are noted rather than one chosen. No attendance figure has been estimated or extrapolated, and events whose 2026 edition had not occurred by 19 August 2026 are shown without a 2026 figure.

Six gaps we could not close, stated because a report that hides them is worth less than one that does not:

• No public dataset produces a clean industry-wide attendance index for tech and crypto conferences.
• No standardised series exists for conference-week hotel rates against normal-week rates across the major host cities, so the cost hypothesis can be evidenced by exhibitor sentiment but not quantified.
• No dataset links visa refusals to conference no-shows, despite the claim being widespread.
• No deal value was disclosed for Hyve's acquisitions.
• No defensible time series exists for conflict-related event-insurance premiums, though war and terrorism exclusions in standard policies are well documented.
• The exhibitor cost research is robust but dates from 2023, so it should not be read as current pricing.

Where this says something is not established, it means we looked and did not find it. It does not mean it is false.

Aspire organises its own events and does not organise any of the conferences discussed here. Dates, prices and attendance for all of them are maintained in our conference directory, with the date each entry was last checked printed on it, and the whole calendar sits in one table.