Mitsubishi Heavy Industries recorded a ¥67.5 billion product warranty provision in Q2 FY2026 tied to its Nakoso and Hirono IGCC plants, generating an operating loss [2][1]. The disclosure lands at a critical moment: AI data center power demand is projected to more than double to 945 TWh by 2030 [3], and the industry is pivoting hard toward on-site generation to bridge a structural grid gap [3]. MHI's loss is a direct warning about the financial risks embedded in the advanced energy infrastructure that hyperscalers and colocation operators now depend on.
What is Covered in this Article
- MHI's ¥67.5B IGCC warranty provision and operating loss [2][1]
- AI data center power demand growth to 945 TWh by 2030 [3]
- Structural mismatch between data center and grid build timelines [3]
- On-site power generation as a permanent infrastructure strategy [3]
- Gas turbine supply constraints and delivery risk [3]
- Debt-funded hyperscaler capex and vendor execution tolerance [3]
The News: On November 6, 2026, Mitsubishi Heavy Industries announced it recorded an operating loss related to product warranty expenses for the Nakoso and Hirono Integrated Coal Gasification Combined Cycle plants [1]. The company accrued a product warranty provision of ¥67.5 billion in Q2 of the fiscal year ending March 31, 2027, resulting in an operating loss [2][1]. MHI classified the charge under Management and Finance [1]. The Nakoso and Hirono IGCC plants represent advanced coal gasification technology, and the scale of the provision reflects the deep technical complexity of delivering such infrastructure at commercial scale. MHI Group spans energy, smart infrastructure, industrial machinery, aerospace and defense, and describes itself as one of the world's leading industrial groups [1].
MHI's ¥67.5B IGCC Loss Flags Execution Risk in AI's Power Bet
Analyst Take: MHI's ¥67.5 billion warranty charge is more than a balance sheet event [2]. It surfaces the execution and financial risks that accompany advanced energy infrastructure delivery at exactly the moment the global infrastructure market is betting on that same class of technology to power the AI era. The timing makes this disclosure a signal worth reading carefully across the hybrid cloud and infrastructure sector.
The Power Gap Driving On-Site Generation Demand
The structural context behind MHI's loss matters as much as the loss itself. The IEA estimated that data centers consumed approximately 415 TWh of electricity globally in 2024, around 1.5% of total global consumption, and projects this figure will more than double to 945 TWh by 2030, roughly equivalent to Japan's current total annual electricity use [3]. The core problem is timing: data centers can be built in 12 to 18 months, while new grid-connected power generation takes between three and seven years to come online [3]. That gap is not closable through grid procurement alone. It is forcing hyperscalers and colocation operators toward a structural, permanent commitment to on-site generation, precisely the domain where MHI competes.
On-Site Power Becomes a Permanent Strategy
The shift to on-site generation is no longer a contingency plan. Data center industry professionals expect 33% to operate on 100% on-site power by 2030, according to a Bloom Energy survey [3]. That figure implies a large and growing market for the kind of advanced power generation systems MHI designs and delivers. It also raises the stakes on execution quality. When one-third of the industry's power supply depends on vendor-delivered on-site systems, warranty failures and delivery shortfalls carry direct operational consequences for the buyer. MHI's IGCC provision quantifies what those consequences can look like from the vendor side [2].
Supply Constraints Amplify Delivery Risk
MHI's disclosure arrives in an already-constrained supply environment. Surging orders for gas turbines have hit a global supply chain that had limited investment for years, and leading manufacturers now report backlogs that can delay the commissioning of new gas plants beyond 2030 [3]. In that context, warranty and reliability performance becomes a critical differentiator. Buyers cannot easily switch vendors or accelerate alternative timelines when backlogs stretch across the decade. A warranty failure of this scale signals that even established industrial leaders face meaningful execution risk on complex energy projects. For infrastructure buyers evaluating long-term generation partnerships, MHI's disclosure reinforces that contract structures must account for delivery and performance risk explicitly.
Financial Risk for Debt-Funded Infrastructure Buyers
The financial environment amplifies the concern. Capex for the hyperscaler group now exceeds internal cash generation, and Morgan Stanley and JP Morgan project the sector may need to issue up to $1.5 trillion in new debt over the coming years [3]. Debt-funded buyers have limited tolerance for vendor-side warranty failures that extend ROI timelines or require unplanned remediation spend. The Asia-Pacific region, MHI's home market, is also the fastest-growing geography for hybrid cloud and infrastructure investment, with the market projected at $32.4 billion and growing at 24.7% [4]. Regional execution credibility matters more, not less, as that investment accelerates.
What This Means for Vendor Selection and Contract Structures
MHI's ¥67.5 billion provision is a cautionary data point for the broader infrastructure market [2]. It does not disqualify advanced energy infrastructure as a solution to the AI power gap. The structural demand is real and growing [3]. But it does reinforce that energy infrastructure delivery is a project execution challenge with direct financial consequences, not just a procurement decision. Hyperscalers and colocation operators evaluating on-site generation partnerships should expect to see warranty terms, performance guarantees, and delivery milestones become central negotiating points. The energy supply constraint shaping infrastructure investment through the end of the decade is as much a financial and execution risk as it is a capacity challenge.
What to Watch
- MHI remediation timeline: whether the company discloses a resolution path for the Nakoso and Hirono IGCC plants in its Q3 FY2026 results [1]
- Hyperscaler contract terms: how on-site generation agreements evolve to include performance bonds, warranty caps, or phased commissioning requirements in Q1 2027 and beyond
- Gas turbine backlog movement: whether leading manufacturers report any relief in order backlogs or delivery timelines in their next earnings cycles [3]
- SMR commercialization milestones: whether small modular reactor projects announce binding offtake agreements or regulatory approvals that could offer an alternative to gas-based on-site generation before 2028
- Asia-Pacific infrastructure investment: how the region's 24.7% growth trajectory in hybrid cloud and infrastructure affects demand for locally sourced energy generation capacity [4]
Sources
1. Notice of Loss Recognition Related to Provision for …, MHI, November 2026
2. Web Source, Jsignal
3. AI Grid Constraints Will Push Over 33% of Data Centers Off-Grid by 2030, Futurum Research, March 2026
4. 2H 2026 Hybrid Cloud & Infrastructure Market Sizing & Five-Year Forecast, Futurum Research, October 2026
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Read the full Futurum Group Disclosure.
Other Insights from Futurum:
Mitsubishi Heavy Industries AI Infrastructure Shift
Author Information
This content is written by a commercial general-purpose language model (LLM) along with the Futurum Intelligence Platform, and has not been curated or reviewed by editors. Due to the inherent limitations in using AI tools, please consider the probability of error. The accuracy, completeness, or timeliness of this content cannot be guaranteed. It is generated on the date indicated at the top of the page, based on the content available, and it may be automatically updated as new content becomes available. The content does not consider any other information or perform any independent analysis.
