May 20, Beijing. At the IDC Directions 2026 forum, a figure left many CEOs and CIOs in the audience silent: by 2026, 50% of enterprise AI projects will fail to meet ROI targets.
On the same day, Huawei released the white paper "Intelligent World 2035" at another closed-door meeting, presenting a different picture: over the next decade, the computing power of the entire society will increase by 100,000 times, and spending on embodied intelligence will surge from $1.4 billion to $77 billion within five years.
Looking at these two things together, it's quite contradictory, isn't it? On one hand, it says the opportunities are unimaginably huge; on the other, it says half the projects will fail. I carefully compared the two reports and found they are not contradictory—they are two sides of the same coin.
$940 billion global enterprise AI spending in 2026 (IDC) $2.1 trillion forecast for 2029, increasing 2.2 times in four years
Two reports, one judgment: the question of keeping up
First, look at the numbers. IDC's data granularity is very fine, broken down into several lines:
- Enterprise AI spending growth rate: $940 billion in 2026, $2.1 trillion in 2029 — a 2.2-fold increase in four years, with a compound annual growth rate of approximately 22%. This pace would be considered fast in any technology track, but in the AI field, it may be just the beginning.
- MaaS (Model as a Service): In 2026, token calls reach 400 trillion times, with market revenue of approximately 18.6 billion RMB. Large models are transitioning from "showing off muscles" to infrastructure-level priced commodities.
- Embodied Intelligence: Currently $1.4 billion, projected to reach $77 billion in five years, with a compound annual growth rate exceeding 120%. This is the only track described as "explosive" in both reports simultaneously.
Huawei's vision is even grander. By 2035, the total computing power of society will increase by 100,000 times (from 10^18 to 10^22 FLOPS), the robot density will rise from 300 to over 1,000 units per 10,000 people, and the manufacturing defect rate will be reduced to below 0.05%. Their logic is: AI moves from "generation" to "action," and the physical world is the main battlefield.
But Huawei's report also has a very interesting blank — completely failing to mention the failure rate.
IDC's Warning: Four Ways AI Can Die
IDC has broken down the reasons for AI project failures into four pitfalls:
| Failure Reason | Key Data | My understanding |
|---|---|---|
| Unclear returns | Projects without a clear ROI path have a failure rate exceeding 70%. | It’s not that the technology isn’t good enough; it’s that they haven’t clearly figured out what problem to solve. Many projects start with just saying "we need to adopt AI," but when it comes to how much money it actually saved or how much efficiency it improved after adoption, they can’t say. |
| Weak data foundation | 60% of enterprises are in a state of decentralized AI management | Data silos, poor quality, and lack of unified governance. This is almost a replica of all ERP implementation projects—without master data governance, upper-level intelligence is nothing but a castle in the air. |
| Human-machine collaboration is weak | 40% of enterprises will shift to unified governance by 2027 | The organization is not ready, and employees either cannot or are unwilling to use it. Purchasing tools does not mean they are in place. |
| Each acts on its own. | By 2027, 40% of enterprises will replace decentralized management with unified governance. | Various departments are conducting their own AI pilot projects without unified data standards, security boundaries, or evaluation mechanisms. It's chaotic. |
Here is a number worth considering: IDC says 60% of enterprises are still in a state of decentralized AI management, but simultaneously predicts that by 2027, 40% of enterprises will shift to unified governance. This means that in the next 18 months, a large number of enterprise CIOs will need to turn "AI governance" from a PowerPoint slide into an organizational structure.
Huawei sees the direction, IDC sees the pitfalls
I made a comparison:
| Dimension | Huawei "Intelligent World 2035" | IDC Directions 2026 |
|---|---|---|
| Perspective | Technology Vision (2035) | Current landing (2026) |
| Risk of Failure | Completely unmentioned | Clear warning: 50% of projects fail |
| Embodied Intelligence | Emphasize technological breakthroughs | Emphasize speed of market adoption |
| Industry AI | The technical path is clear | Data foundation is the biggest bottleneck |
| Core Narrative | How big is the chance | How many people will die on the road |
This comparison is not to say that Huawei's report is problematic—they are inherently focused on technological vision. But for corporate CIOs, having both reports together provides a complete decision-making map: The direction is right, but the road is full of pitfalls.
Three Practical Suggestions for Enterprise CIOs
Based on these two reports and the actual situation of the enterprises we have contacted, I have identified three points:
First, manage your data well before talking about AI. Two out of the four types of failures are directly related to data governance. Among the manufacturing AI projects I've seen, six out of eight get stuck on data quality—either the material codes in the ERP system are inconsistent, or the equipment data collection is incomplete. The 100,000-fold computing power Huawei talks about sounds far off, but if you don't manage your master data now, even if that 100,000-fold computing power arrives, you won't be able to use it.
Second, clarify the ROI before taking action. IDC says that over 70% of projects without a clear ROI path fail, and that's not an exaggeration. In the enterprise AI research I conducted last year, fewer than 20% of companies could clearly state "how much money this AI project saved in a year." Most answers were "improved efficiency" or "enhanced experience." It's not that these statements are wrong, but they cannot be verified. Things that cannot be verified will inevitably have their budgets cut.
Third, don't wait. Decentralized management is the reality; you don't need to achieve unified governance in one step. You can first pilot it in one department—for example, using AI assistants to handle financial reimbursement audits, or using agents for customer service script quality checks—calculate the ROI, and then scale it. This approach is far more practical than setting up a "Group-wide AI Governance Committee" to issue a couple of documents.
Huawei says the AI super cycle has arrived, while IDC says 50% of projects will fail. Both statements hold true. The difference lies in whether you are the former or the latter.
Source: IDC Directions 2026 Beijing Forum (May 20, 2026), Huawei "Intelligent World 2035" White Paper
Disclaimer: This article represents only the author's views and does not constitute investment advice. Reproduction must indicate the source.
