Introduction

Three AI-driven infrastructure companies Vertiv Applied Optoelectronics and Innodata have all retreated from their 2026 peaks prompting investors to evaluate whether the decline offers a entry point or reflects underlying concerns While each operates at the intersection of artificial intelligence and essential hardware their financial health growth drivers and risk exposures vary considerably

What Happened

Vertiv reported second-quarter revenue of $3.27 billion a 24 percent increase with adjusted operating margin expanding 410 basis points to 22.6 percent and adjusted free cash flow hitting $925 million Despite these results the stock trades well below its recent highs as 112 hedge funds held the name in Q2 up from 96 in Q1 Applied Optoelectronics delivered record quarterly revenue of $191.9 million fueled by more than double the 800G optical transceiver volume but the company posted a $22.8 million GAAP loss and a 27.7 percent gross margin Fifty-four hedge funds remain invested down from 55 while Value Aligned Research increased its stake by 26 percent sequentially Innodata revenue rose 58 percent to $92.1 million and adjusted EBITDA climbed 92 percent to $25.4 million supported by reusable data sets and evaluation services However the company depends heavily on a few large customers and faces rapid shifts in AI training methodologies Hedge fund participation grew to 25 from 20 in the latest quarter

Why This Matters

The pullbacks across these three AI-related names are not interchangeable Vertiv risk hinges on project timing and supply-chain congestion in data-center construction Applied Optoelectronics outlook depends on manufacturing ramp and customer concentration while Innodata exposure comes from customer concentration and evolving model-training practices Elevated short interest in Applied Optoelectronics with approximately 10.4 million shares sold short and a low days-to-cover ratio adds sentiment pressure Investors must assess whether each companys fundamentals can support the valuations implied by their current price levels

Key Takeaways

  • Vertiv exhibits the strongest cash generation of the group but its valuation hinges on delayed data-center construction timelines
  • Applied Optoelectronics offers the most significant capacity upside with 800G demand expected to outstrip production through mid-2027 though near-term margins and execution remain uncertain
  • Innodata has the lightest physical footprint and the fastest revenue growth yet large-customer dependence and shifting training methods present meaningful risk
  • A buying opportunity only materializes if margin durability customer diversification and cash conversion justify each stocks current price
  • Order conversion gross-margin stability customer concentration and diluted share growth serve as the most comparable checkpoints when assessing entry points

Conclusion

While the pullback from recent highs improves entry prices for each name operating evidence not simply distance from a peak must justify investment decisions Balance-sheet resilience margin sustainability and customer diversification will likely determine which if any of these AI-linked stocks deliver compelling risk-adjusted returns Investors should avoid relying on past highs as a valuation shortcut and instead focus on concrete progress in order flow margin protection and shareholder-friendly cash generation