Корпоративный AI4 мин. чтения

IBM in Crisis: Mainframe Sales Plummet 42% Amid AI Boom

IBM reports worst quarter in a decade with 42% mainframe sales collapse. Here's why financial institutions are abandoning legacy systems for AI infrastructure.

IBM in Crisis: Mainframe Sales Plummet 42% Amid AI Boom

July 22, 2026 – IBM stunned investors with $17.2B revenue (-15% vs forecast) as shares plunged 25%. The culprit? A 42% nosedive in mainframe sales, which generate 35% of corporate profits. CEO Arvind Krishna calls this a temporary budget shift, but data reveals deeper structural issues.

IBM mainframe sales collapse chart

At a glance: Key facts

  • Historic drop: IBM mainframes post 42% quarterly sales decline
  • Domino effect: Every $1 in hardware drives $3 in software revenue
  • Pricing shock: Server components rose 15-30% due to AI demand
  • Market reaction: IBM stock crashes 25% in single day
  • Survival strategy: Hybrid solutions with AI accelerators
  • Regional impact: Asia-Pacific suffers worst losses (-51%)
  • Recovery timeline: Analysts predict 4+ quarters of stagnation

Financial freefall: The numbers tell the story

IBM's Infrastructure segment (home to mainframes) posted its worst performance in a decade. Despite 58% margins, clients are freezing contracts en masse. Three core issues:

  1. Component costs: NVIDIA H100 GPUs up 32%, HBM3 memory up 28%
  2. Budget competition: 60% of deferred deals come from banks shifting to cloud AI
  3. Sales cycle lag: 7-10 year mainframe lifespan slows market response

Additional pressure points:

  • Refresh rates drop: Only 18% upgraded systems in 2026 vs 25% in 2025
  • Support cost hike: Maintenance contracts rose 11% on average
  • Skills shortage: Certified z/OS engineers down 7% YoY

AI vs mainframes: The budget battle

Forrester Research confirms companies are sacrificing legacy upgrades for AI infrastructure. Cost comparison:

Resource Price increase Impact
GPUs +32% IT budget reallocation
Cloud AI services +18% Reduced mainframe demand
Power consumption +22% Higher mainframe TCO

Case study: JPMorgan tests IBM's hybrid solution, but 12 rival banks already moved 15% workloads to AWS/Azure. CitiGroup internal documents show 28% cost savings on transactions using cloud AI.

Crisis depth: Hidden consequences

IBM contract analysis reveals alarming trends:

  • Average deal size fell from $4.2M to $2.8M
  • 27% clients shortened service terms from 5 to 3 years
  • New mainframe clients just 8% vs 15% in 2025

Additional risks:

  • Decision cycles doubled to 6 months for large deals
  • Competition heats up: Dell/HPE offer 20% migration bonuses
  • Supply chain delays: Telum chips for z16 face 14-week backlogs

IBM's counterattack: Three strategic bets

  1. z16 AI Accelerator – Dedicated chip for mainframe AI workloads (claims 40% savings)
  2. Quantum encryption – Play to retain banking clients
  3. Cloud integration – AWS/Azure APIs instead of direct competition

z16 AI Accelerator specs:

  • TensorFlow/PyTorch compatibility
  • 16TB RAM per node
  • 35% lower transaction latency
  • Quantum attack protection
  • Hybrid cloud GPU integration

Outlook: From optimism to catastrophe

  • Realists: Mainframe market stabilizes at $3.5B/year
  • Pessimists: IBM loses 25% clients by 2028
  • Skeptics: New IT talent avoids legacy systems

IDC survey shows 43% CIOs view mainframes as "stopgap until 2030." Yet 67% note full migration would take 5-7 years.

Alternative scenarios

  • M&A: IBM may acquire niche AI startups
  • Restructuring: Spin off mainframe business unit
  • Radical shift: Subscription licensing model

Questions & answers

Why do mainframes still matter?

Banks can't fully migrate transactional systems due to security and legacy code. Visa tests show processing 10,000 transactions/sec is 27% cheaper on mainframes. Additional advantages:

  • Five-nines reliability (99.999% uptime)
  • Regulatory compliance: Some data can't be cloud-hosted
  • Migration complexity: Rewriting COBOL apps takes years

What are IBM's biggest risks?

  • Losing financial sector (60% of segment revenue)
  • Client defections to Oracle/Microsoft hybrid solutions
  • z/OS talent shortage – Only 12% universities teach COBOL
  • Technical debt: 40% of mainframe code needs modernization
  • Energy regulations: New efficiency standards may raise costs

Can IBM recover?

Yes, but only if hybrid strategy succeeds. JPMorgan tests show 40% savings – a potential lifeline. However, Morgan Stanley estimates 3-5 years to pre-crisis levels. Key success factors:

  • Faster AI tool integration
  • 25%+ TCO reduction by 2027
  • Expanding partner ecosystem

How does AI affect ownership costs?

IBM's new AI modules promise:

  • Auto-scaling resources
  • 22% power savings
  • Predictive maintenance
  • Real-time workload optimization
  • 30% staff cost reduction via automation

What's the talent pipeline outlook?

IBM launched three initiatives:

  1. University grants for COBOL curriculum
  2. Cloud engineer retraining
  3. EdTech partnerships for online training