I Walked the AWS Summit London 2026 Expo Floor. These 6 Words Told Me Everything About Where Enterprise Tech Is Heading.
AWS Summit London 2026: 6 Words That Signal Where Enterprise Is Heading | PivortalHub
I was at AWS Summit London 2026 at ExCeL yesterday — 7,000+ attendees, 200+ sessions, the largest cloud technology gathering in the UK calendar. I spent most of my time not in the breakout rooms but on the expo floor, doing something I have never seen written about but which I believe is one of the most reliable ways to read where enterprise demand is actually heading.
Forget the keynotes. Forget the analyst briefings. Forget the carefully curated session abstracts.
Read the booths.
Every word on an expo stand has been through a product team, a marketing team and often a C-suite review. The companies paying tens of thousands of pounds for a floor presence at AWS Summit London have made a calculated commercial bet that the words above their stand will stop the right people in their tracks. Those words are not aspirational. They are market signals. They represent where the vendor believes enterprise money will flow over the next twelve to eighteen months.
7,000+
Attendees at AWS Summit London 2026 — ExCeL, Royal Victoria Dock, 22 April
200+
Sessions spanning agentic AI, security, data and digital transformation
6
Words that defined the expo floor — and what each one tells you about enterprise demand
Six words defined the AWS Summit London 2026 expo floor. Here is what each one told me about the state of enterprise technology — and what it means for delivery managers and FinOps practitioners.
1
Telemetry
Datadog · Dynatrace · Tsuga · multiple platforms
What the word signals about market demand
There is a data crisis hiding in plain sight inside most enterprise technology organisations. Omdia's January 2026 research found that 55% of business leaders say they lack the data to make effective decisions about technology spending. Not lack of dashboards. Not lack of tools. Lack of the right signal at the right moment.
Here is the paradox: organisations have never had more data. LogicMonitor's 2026 observability research found that 59% of organisations are drowning in telemetry but cannot get answers when they need them. They can see that something broke. They just cannot figure out what, why or how to fix it — let alone prevent it next time.
The vendors using "telemetry" on their booths are not selling data collection. They are selling signal extraction — the ability to pull the meaningful signal from exponentially growing noise.
Here is the paradox: organisations have never had more data. LogicMonitor's 2026 observability research found that 59% of organisations are drowning in telemetry but cannot get answers when they need them. They can see that something broke. They just cannot figure out what, why or how to fix it — let alone prevent it next time.
The vendors using "telemetry" on their booths are not selling data collection. They are selling signal extraction — the ability to pull the meaningful signal from exponentially growing noise.
What this means for delivery managers
Telemetry is no longer just an infrastructure concern. AI sprint teams are generating telemetry — cost per inference, model response time, API call frequency — that nobody on your sprint board is currently monitoring. The Definition of Done for any AI feature should include: what telemetry will this feature emit, who will watch it, and what threshold triggers a review?
2
Observability
Tsuga · Datadog · Dynatrace · Splunk
What the word signals about market demand
The market has moved through three phases. Monitoring — knowing something is wrong. Observability — understanding why. And now in 2026, the third phase is emerging: autonomous remediation — fixing it without human intervention.
96% of organisations are maintaining or increasing observability spending. 84% are pursuing unified platforms to reduce tool complexity. But the more important signal is this: Elastic's 2026 research confirms that governance, risk and compliance teams are now adopting observability data in 43% of organisations — extending its remit from "is the system running" to "is the system behaving within policy."
Observability has become the foundation of governance. And governance, in 2026, is the dominant enterprise concern.
96% of organisations are maintaining or increasing observability spending. 84% are pursuing unified platforms to reduce tool complexity. But the more important signal is this: Elastic's 2026 research confirms that governance, risk and compliance teams are now adopting observability data in 43% of organisations — extending its remit from "is the system running" to "is the system behaving within policy."
Observability has become the foundation of governance. And governance, in 2026, is the dominant enterprise concern.
What this means for delivery managers
Flow efficiency — the ratio of active work time to total elapsed time — is an observability problem applied to delivery. The teams that instrument their delivery pipeline with the same rigour they apply to their infrastructure will find the waste, the bottlenecks and the cost of delay that is currently invisible.
3
Agentic
Finout · TCS · AWS keynote theme
What the word signals about market demand
Gartner predicts 40% of enterprise applications will embed AI agents by the end of 2026 — up from less than 5% in 2025. The agentic AI market is projected to grow from $7.8 billion to over $52 billion by 2030.
What makes "agentic" different from "AI" as a booth word is the implication of autonomy. Agentic AI systems do not wait to be asked. They reason, plan, act and adapt continuously — spinning up resources, calling APIs and accumulating costs without direct human instruction for each action.
Finout's positioning — "FinOps for the Agentic Era" — is the most forward-looking claim I saw at the Summit. They are explicitly building for the governance problem that most organisations have not yet encountered but will encounter within the next twelve months.
What makes "agentic" different from "AI" as a booth word is the implication of autonomy. Agentic AI systems do not wait to be asked. They reason, plan, act and adapt continuously — spinning up resources, calling APIs and accumulating costs without direct human instruction for each action.
Finout's positioning — "FinOps for the Agentic Era" — is the most forward-looking claim I saw at the Summit. They are explicitly building for the governance problem that most organisations have not yet encountered but will encounter within the next twelve months.
What this means for delivery managers
The sprint board of 2027 will include AI agent stories. Those stories will have cost profiles — cost per run, cost per decision, cost per outcome — that need to be estimated, monitored and governed. The delivery manager who understands agentic AI economics before their product teams start deploying agents will be the most valuable person in the room.
4
Autonomous
ProsperOps · infrastructure platforms
What the word signals about market demand
There is a critical distinction between automated and autonomous that booth copy at AWS Summit London was making explicit. Automated means a human defined the rules and the machine follows them. Autonomous means the machine makes the decisions within defined parameters — without human approval for each action.
ProsperOps — managing $6 billion in cloud spend and having returned $4 billion in savings to customers — used "autonomous" deliberately. They execute Reserved Instance commitments. Not recommend them. Execute them. Flexible commitments that adapt as usage evolves, without human sign-off on each transaction.
But Gartner adds an important counterweight: more than 40% of agent projects will fail by 2027 due to runaway costs, unclear business value and agents that violate policy or create risk.
ProsperOps — managing $6 billion in cloud spend and having returned $4 billion in savings to customers — used "autonomous" deliberately. They execute Reserved Instance commitments. Not recommend them. Execute them. Flexible commitments that adapt as usage evolves, without human sign-off on each transaction.
But Gartner adds an important counterweight: more than 40% of agent projects will fail by 2027 due to runaway costs, unclear business value and agents that violate policy or create risk.
What this means for delivery managers
Every autonomous system requires a governance boundary. The Andon Cord principle — giving any team member the authority to stop the line when a problem is detected — applies directly to autonomous AI. Who has the authority to halt an autonomous agent? What triggers escalation? What is the Definition of Done for a safely deployed autonomous system? These are delivery questions, not technology questions.
5
Trusted
WSO2 · security vendors throughout
What the word signals about market demand
The word "trusted" at a technology conference in 2026 is doing enormous commercial work. WSO2's positioning — "Trusted AI Governance" on AWS Marketplace — is a direct response to a fear that every board, every regulator and every CTO is now navigating: can we deploy this AI with confidence?
Elastic's 2026 observability survey found that 61% of organisations cite security as the primary concern when implementing generative AI. Gartner's analysis confirms that agents operating with autonomy create constant risk — bad data handling, policy violations and unintended actions are all documented failure modes.
The critical nuance: "trusted" is not a technical property. It is a governance property. You cannot make an AI system trusted by writing better code. You make it trusted by building accountability structures, audit trails, human-in-the-loop oversight and policy guardrails.
Elastic's 2026 observability survey found that 61% of organisations cite security as the primary concern when implementing generative AI. Gartner's analysis confirms that agents operating with autonomy create constant risk — bad data handling, policy violations and unintended actions are all documented failure modes.
The critical nuance: "trusted" is not a technical property. It is a governance property. You cannot make an AI system trusted by writing better code. You make it trusted by building accountability structures, audit trails, human-in-the-loop oversight and policy guardrails.
What this means for delivery managers
Trusted AI is not a feature. It is a characteristic of how a team builds, tests, deploys and monitors AI systems continuously. The Definition of Done for any AI feature needs a trust criterion — not just "does it work" but "does it behave within policy, with full telemetry, with a clear escalation path when it does not."
6
Beyond Cloud
FinOps+ by Kion · multi-technology platforms
What the word signals about market demand
Two words that tell you the cloud cost management market is saturated. If you still lead with "we do cloud cost optimisation" in 2026, you are behind. The vendors who have won that category are now positioning for what comes next.
"Beyond Cloud" means: SaaS spend (90% of organisations). AI spend (98%). Software licensing (64%). Data centre (48%). The scope of technology value management has expanded beyond what any cloud-native FinOps tool was designed to handle.
The FinOps Foundation's 2026 mission update — from "value of cloud" to "value of technology" — is the policy statement for exactly this shift. FinOps+ by Kion's booth copy at AWS Summit London is the market's commercial response.
"Beyond Cloud" means: SaaS spend (90% of organisations). AI spend (98%). Software licensing (64%). Data centre (48%). The scope of technology value management has expanded beyond what any cloud-native FinOps tool was designed to handle.
The FinOps Foundation's 2026 mission update — from "value of cloud" to "value of technology" — is the policy statement for exactly this shift. FinOps+ by Kion's booth copy at AWS Summit London is the market's commercial response.
What this means for delivery managers
The conversation your CTO needs to have about technology spend is not "what are we spending on AWS." It is "what is the total cost of our technology portfolio across cloud, SaaS, AI, licensing and data centre — and what is each pound of that spend actually delivering?" That conversation requires the Shift Up capability. And no vendor booth can have it for you.
The meta-framework: why booth copy is the most honest market signal available
Let me make the framework explicit because I think it is genuinely useful for anyone trying to read where enterprise technology is heading.
Analyst reports are lagging indicators. They describe what organisations have already adopted. Vendor press releases are promotional. Conference session titles are curated.
Booth copy is different. It is written under commercial pressure to stop a specific person — the CTO, the cloud architect, the delivery manager with a budget problem — in their tracks within three seconds. Every word is tested against the question: will the right person stop walking when they see this?
The gap nobody is selling
Every vendor at the Summit is building tools. Nobody is building the culture change that makes those tools work.
The CloudZero dashboard does not govern itself. The ProsperOps autonomous engine does not coach teams to care about cost. The Finout agentic era platform does not run the Scrum of Scrums that surfaces AI spend as a delivery risk before it becomes a billing shock.
Someone has to embed cost attribution into the Definition of Done. Someone has to implement the Andon Cord so ungoverned AI usage is flagged at the point of architectural decision. Someone has to shift teams from feature delivery thinking to product ownership — from "we built what we were told" to "we own the outcome and its cost."
That someone is not a SaaS platform. That someone is a delivery manager who understands both the technology and the people — and has the coaching experience to change how teams think about value, not just how they process tickets.
The tools are getting better. The culture gap is getting wider. AWS Summit London 2026 confirmed both.
Someone has to embed cost attribution into the Definition of Done. Someone has to implement the Andon Cord so ungoverned AI usage is flagged at the point of architectural decision. Someone has to shift teams from feature delivery thinking to product ownership — from "we built what we were told" to "we own the outcome and its cost."
That someone is not a SaaS platform. That someone is a delivery manager who understands both the technology and the people — and has the coaching experience to change how teams think about value, not just how they process tickets.
The tools are getting better. The culture gap is getting wider. AWS Summit London 2026 confirmed both.
The Delivery Brief · Monthly Newsletter
Practitioner thinking on FinOps, AI and Agile — direct to your inbox.
No vendor content. No recycled frameworks. One monthly issue written from inside live programmes, client engagements and — apparently — expo floors at ExCeL London.
Subscribe free →
What word on an expo booth has caught your eye recently — and what do you think it signals about market demand? Drop it in the comments. The most interesting conversations on this platform happen when practitioners push back on the framework.
Further reading: What AWS Summit London 2026 told me about the future of FinOps · IBM Observability Trends 2026 · State of FinOps 2026
← Previous article
What AWS Summit London 2026 Told Me About the Future of FinOps
IJ
Issouf Jilany
FinOps & Cloud Cost Optimisation Consultant · SAFe SPC · SAFe RTE · AWS Solutions Architect · IBM Apptio Cloudability · FOCP (In Progress)
Senior FinOps practitioner and Cloud Cost Optimisation Consultant with 20 years of experience across Lehman Brothers, Reuters, Lloyd's of London, HMRC and OFGEM. Attended AWS Summit London 2026 as a practitioner — not a vendor. Founder of PivortalHub — writing practitioner-led frameworks on FinOps, AI adoption and Agile delivery at pivortalhub.co.uk.
