Enterprise hits and misses - CIOs contrast AI results versus reality, while the Block 'AI washing' debate erupts

Lead story - CIOs navigate AI - Why 93% of enterprises use AI but most aren't seeing the ROI they were promised

The markets want to argue about AI versus SaaS distruptions, but things are different on the ground. For customers, AI adoption versus ROI is poised to be the hot enterprise conversation of the year.

If you haven't tracked the data from the diginomica network on this topic, now is the time. Recently, we published insights from direct conversations with 35 CIOs and CTOs, who collectively oversee billions in technology spend (you can see the data highlights, and sign up for the full free report). 

The top finding? The continued disconnect between AI hype and implementation reality. This maps with my own stump speech: vendors are getting ahead of customers on AI sophistication; that gap puts projects in peril. On the event trail, customers routinely tell me: "I liked what I heard, but I don't know how to get there."  But what did the diginomica network findings say? 

The expectation gap is real. While 93% of organizations now use AI in some form, only 21.4% report success rates above 80%. High-profile implementations — including Microsoft Copilot, automated bid tools, and headcount reduction initiatives — are delivering disappointing returns.

Crucial point: these tech leaders indicate that AI technology itself is not the barrier.

The barriers to AI value realization are organizational, not technical. Poor data quality, disconnected data systems, and inadequate change management top the list of blockers. Organizations are repeating the pattern that has plagued past technology implementations: seeing only 10-20% of potential benefits because they fail to drive proper adoption and education.

One potentially surprising insight? Investment in tech talent remains a priority. 

When forced to choose between hiring 10 data scientists or buying an enterprise AI platform, 64.3% of technology leaders chose the people. As one CIO explained: 'We need bright people finding bright ways to use bright technology — not technology operating in isolation.'

This all resonates. I do think, however, that while the current state of AI tech is not a barrier to successful results, there are tech barriers with noting - a topic I am admittedly a bit notorious for documenting. This issue usually comes into play via unrealistic expectations, not at the CIO level, but at the board level. I can see similar signs in these diginomica network findings: 

Boards consuming vendor marketing expect transformation; CIOs are managing the nuanced reality on the ground.

Just last week, I talked to a customer who had to walk executives through why AI coding tools like Claude Code and Replit weren't enterprise-grade for their compliance setting (in contrast, other enterprise tools in use had superior data context, governance, and role-based authorizations). 

Since this diginomica network research came out, I've seen Derek contrast these data points with vendor plans, and their own research findings. Rightly so; we can't cross over into consistent AI results until this gap is closed. I'm looking forward to the next wave of community findings. Stay tuned...

Diginomica picks - my top stories on diginomica this week

• Knowledge velocity is the fundamental driver of change today - and human connection is at its heart - Phil has authored one of the essential diginomica think pieces of the year to date. I recommend a full reading... for now, this teaser: "Too often, enterprises look at technology as a means of lowering the cost of what they already do. But this is completely the wrong reaction in the face of rising knowledge velocity."
• If you can’t trust your people, you won’t trust your agents – RTO mandates mean you’re not ready for AI - I've been planning to connect the dots between RTO and AI adoption, but Ian beat me to it, with a forward-thinking post worthy of a good debate... another time.
• CEO mea culpa on show at Ocado as management concedes its early business model is still delivering consequences today - If anyone can make sense of the ups and downs of Ocado's retail automation play, it's Stuart.
• A little bit of history repeating? Déjà vu lessons for AI code development - Brian uncorks some salty/fresh lessons on AI coding, via an unexpected spreadsheets analogy.
• Electrical grids become the biggest AI bottleneck. Here's why - and what might be done about the problem - George bears down on one of the core obstacles to AI infrastructure rollouts - at least in the western hemisphere. 

Vendor analysis, diginomica style. 

With the SaaSpocalypse narrative still running (too) hot across stock markets, we have two earnings reports of note to consider: 

• Next chapter please! Workday CEO Aneel Bhusri reads the future and shoots down the 'SaaSpocalypse' - Stuart on Workday's good (but not good enough for the short termers) year, and Bhusri's measured response. But it's Bhusri's offhand quip that deserves the pull quote: "Every one of these AI leaders actually runs Workday just for what it's worth - Anthropic, Google and OpenAI all run Workday." That strikes me as a crucial proof point...
• How much is that worth in AWUs? Introducing a new metric to track agentic value as Salesforce turns in strong full year numbers - Again strong numbers that didn't sway markets, but as Stuart reports, that didn't stop Benioff from going off. Salesforce is consuming enough AI tokens to chart on the top worldwide LLM leaders... but quoting Benioff, that's not necessarily a proxy for value. Thus the AWU news: "But we really want to take this to another level, and another level is a token on its own doesn't know your customers, your pipeline, your org chart, but Salesforce does. The value isn't in the token, the value is what our platform does with it. That's why today we're introducing the Agentic Work Unit (AWU)." Now that will be interesting to track...

Meanwhile, here's my top choices from our vendor coverage:

• When healthcare admin slows down, patients wait. UiPath's agentic AI is designed to change that - Alyx has one of the most compelling takes on AI and healthcare I've seen this year. Yes, let's hone in on problems that change patient care: "The critical question is not whether tasks can be completed faster. It is whether productivity gains translate into structural change: fewer denials, shorter turnaround times, improved first-pass accuracy and expanded capacity without corresponding cost growth."
• Slashed headcount, hefty expenses cuts - can C3.ai swallow its own AI medicine to clean up its current organizational mess? C3.ai is one of the most fascinating enterprise AI vendors to track, but it's been a rocky road. Stuart's on the case again...
• Agentic AI – you’ll see million-dollar losses before million-dollar gains, warns Cloudera CTO - Chris drills into Cloudera's take on getting AI agents right, and where the pitfalls still lie.
• ServiceNow's Autonomous Workforce is here and it's impressive - are enterprises ready for it? - Derek's take on ServiceNow's latest agentic moves sets up what is sure to be a fascinating ServiceNow KnowledgeNow in Vegas later this spring. 

A few more vendor picks, without the quotables:

• Fibr.ai wants to re-invent CRO with an agentic web experience layer. Here’s why it matters - Barb
• How Accenture is minding the AI enterprise adoption gap - Katy
• Canva adds video to its pro design toolset and steps up its enterprise play - Phil
• SAP and the 'SaaSpocalypse' - damaging AI displacement won't get past the defensive moats, argues Chief AI Strategy Officer Sean Kask - Stuart
• Everpure posts first $1 billion revenue quarter - the numbers are making the case for the rebrand - Derek

Jon's grab bag - We're in dire need of some tech-for-good stories this week, and Cath's got a couple of dandies: AML Registry to transform blood cancer research with the help of AI, and How data will enable McCain’s Farm of the Future to boost regenerative farming worldwide. Chris has a scorching book review in Welcome to inverted AI utopia!
Why Dr Vivienne Ming's future looks positively merciless. Gary blows an entertaining gasket going after Anthropic's 'we solved Cobol code' announcement in Monday Morning Moan - messing with our safest IT systems might be the worst thing the AI vandals have come up with...so far!

As the Anthropic/OpenAI/Pentagon story surged, Stuart waded in with frequent missives/deconstructions. If you missed the blow-by-blow, here's the latest: Something for the weekend (and beyond) - Trump 2.0 goes nuclear on "left-wing nut jobs" Anthropic, but the fallout will engulf the entire tech sector, preceded by What Dario Amodei wouldn't do, Sam Altman would! Here's why the OpenAI CEO signed a deal with the US Department of War - and why he wants you to visit him in jail if things go wrong.... Stuart also examined the implications on European tech innovation.

We can conjure a heated argument on these regulatory fisticuffs at any time, but two lines in Stuart's analysis jump out today: "Once the puerile name calling and personal insults die down, we need a lot more transparency on a lot of aspects of what's happened/is happening." And: "We are in uncharted waters here." As for enterprises trying to rationalize their AI pursuits amidst this fracas, I have no quick/easy answers except: plan B is probably on the table, to provide assurances if the vendors in plan A falter. That plan B surely includes expanding the list of trusted partners, to mitigate over-dependencies.

Best of the enterprise web

My top seven

Are the Block layoffs really about AI? 

These stories are never as cut-and-dried as the frantic LinkedIn proclamations. It's too easy to use them as agenda fuel... But as for Block's layoffs being hailed/cautioned as 'truly the first AI-related layoffs,' here is some caution sauce on that media hot cake. Consider Bloomberg's Block’s 4,000 Job Cuts Raise Questions Over AI’s Role in Layoffs. The choice cut? 

Block’s recent history suggests that AI adoption is not the only factor influencing its staffing decisions. The company loaded up on workers during and after the pandemic, more than tripling its employee base between 2019 and 2022, and has been slower than peers to scale back. Its stock had fallen roughly 40% since the beginning of 2025, a trajectory that had nothing to do with AI and everything to do with a business that had grown unwieldy.

Block's peak employee count after the post-pandemic hiring surge? 12,500. Ergo: 

A note from Goldman Sachs published Friday argues that fears of an imminent AI jobs apocalypse are “excessive.” The bank’s economists estimate that sectors like tech that are impacted by AI are shaving just 5,000 to 10,000 jobs per month from overall payroll growth in the US. Goldman forecasts a 0.5 percentage point increase in the unemployment rate as adoption rises.

Did I mention that Block is heavily invested in Bitcoin, to the tune of $1 billion+, a volatile market that is not doing so well of late? And yes, Block was rewarded with a big market bump after the 'AI layoff' news. A sober/thoughtful discussion on AI and jobs is important, even necessary. 

I'm seeing signs of teams punching above their weight with savvy use of these tools, even as Goldman Sachs has indicated the impact of AI on GDP has been, to data, nominal (more on that shortly). Some junior level roles do appear threatened, but other companies (IBM, Walmart) are hiring young talent. Oh, and p.s. - the economy isn't exactly awesome. We could stand to have this important conversation without hyperbole... 

• AI boosted US economy by 'basically zero' in 2025, says Goldman Sachs chief economist — 'We think there's been a lot of misreporting of the impact that AI investment had on GDP growth' - "We don't actually view AI investment as strongly growth positive..." A bit of a contrast with Jack Dorsey's headcount experiments, eh?
• Anthropic vs. SaaS: A nuanced view  - Constellation's Larry Dignan has another thoughtful LLM versus SaaS take, with sharp edges: "In the end, Anthropic is going to help systems-of-record providers yet take away their user interfaces and potentially commoditize them."
• Agents of Chaos paper raises agentic AI questions - Another notable Dignan post: "In several cases, agents reported task completion while the underlying system state contradicted those reports. We also report on some of the failed attempts."
• The real breakthrough in robotics is foundation models — not hardware - via The New Stack, finally a really well thought robotics article - one that looks beyond the less significant impact of LLMs, and into other notable advances.
• The Gift of Feedback - Frank Scavo hits at some home truths of the workplace: "A workplace devoid of honest feedback becomes unhealthy, one with a culture of fear, negativity, and resentment."
• Enterprise month in review - agentic AI gut check time - the replay of our latest no-holds-barred enterprise review, where Brian Sommer and I share our underrated stories of the month, and hash out the surprising agentic insights from Andreas Welsch, via his new book on this topic. More multi-media, if you're so inclined: ZohoDay 2026 - the impromptu podcast recap with Thomas Wieberneit.

Whiffs

So....

If influencers tell you AI is super-fab, does that do the trick? Evidently is does...usually:

Pay an influencer today, replace one with an avatar tomorrow... Meanwhile: 

See you next time... If you find an #ensw piece that qualifies for hits and misses - in a good or bad way - let me know in the comments as Clive (almost) always does. Most Enterprise hits and misses articles are selected from my curated @jonerpnewsfeed.