2Q26 Earnings Review: SaaS, Telco and AI sector performance

The 2Q26 earnings season highlighted stark divergence across key tech and telecom sub-sectors. While artificial intelligence investments continue to drive strong top-line momentum, high capital intensity and rising infrastructure costs are weighing heavily on free cash flow generation. Elsewhere, SaaS issuers face headwinds from deal-cycle scrutiny and churn, while traditional Telecom providers navigate stiff price competition.

In our latest review, we distill the key 2Q26 takeaways and offer essential insight into evolving leverage trends, cash flow conversion, and sub-sector risk factors.

IT Services & Enterprise Software

Conduent, Cvent, Atos and EverCommerce are 4 maturing technology businesses where near-term credit quality is currently being decided by a capital-structure or portfolio event rather than by underlying demand. Cvent carries leverage from debt-funded consolidation (an LBO and an M&A roll-up, respectively), while EverCommerce represents a stable leverage position thru organic deleveraging. Conduent is actively shrinking itself via divestiture to defend margin; Atos split itself into two companies and sold off assets before restructuring its balance sheet directly with creditors: 

Across their most recent periods, management teams describe businesses mid-transformation rather than businesses simply reacting to demand.

Conduent's leadership attributed Q2 2026's weak 3.0% Adjusted EBITDA margin specifically to "lower profitability in our Government segment," while guiding to a materially higher full-year Adjusted EBITDA range of $140-170 million (roughly 6.5-7.5% margin) and disclosing signed divestitures — including its Public Transit and Tolling businesses — expected to generate $234 million in gross proceeds; the message is that the current quarter is a low point the divestiture program and cost actions are meant to correct, even as the company continues working through its previously disclosed January 2025 cyber event.

EverCommerce's newly appointed CEO, Alex Goor, presented a stronger immediate result — Q2 revenue in line with guidance and Adjusted EBITDA of $44.5 million beating the top of its range at a 29.3% margin, with cross-sell adoption up 26% year-over-year — but tempered it by guiding Q3 and full-year 2026 results toward "the lower end" of prior ranges, a cautious signal delivered alongside the leadership change itself.

Atos's 1H26 update is the most explicitly transformation-focused of the three: management reported its three-year "Genesis" cost-savings program already complete ahead of schedule and now being extended, and cited "initial benefits" from a 2025 commercial-organization reset showing up in "productivity and pipeline quality.”  Portfolio simplification continues — the Company completed sale of its Ideal GRP unit in January 2026, the divestiture of its South America operations in April 2026, and a further 20 legal-entity reductions in the half — while naming agentic AI, sovereign computing and cybersecurity as its three forward growth pillars.

On Genesis specifically, management had the following to say in its more recent earnings call:

Phase 1 (completed):

• EUR 650 million in identified gains achieved in Q1 2026
• Completed within one year of programme launch
  

Phase 2 (Ongoing through end of 2026):

• Expanded savings target to EUR 800 million or above
• Additional EUR 100+ million in savings identified for 2027 vs. 2026
• Management is exploring further G&A optimisation and process improvements beyond the traditional Genesis scope, potentially yielding additional savings in 2028

Financial Results & Margin Improvement

• Operating margin increased 43% in 1H26 on a declining turnover, demonstrating Genesis effectiveness
• Group operating margin reached EUR 190 million (~6% EBIT margin on the going-forward perimeter)
• Margin improvement achieved despite a 6.3% inorganic revenue decline

Restructuring Costs were estimated as follows:

2. Telecom & Cable

Every issuer in this group is losing subscribers or revenue — cord-cutting, fixed-line erosion, and mobile/fiber price competition are pressures affecting the entire industry, not any one company — so growth isn't what separates these four credits. What separates them is how actively each is restructuring in response. Lumen sold a legacy consumer fiber business to AT&T for $5.7 billion in cash. Charter recently completed its acquisition of Cox Communications. Optimum is leaning into an AI-infrastructure-adjacent fiber subsidiary even as its core video and broadband business struggles.  Altice France has signed an MoU with a consortium consisting of Bouygues Telecom, Orange and Iliad to sell SFR.

Altice France and Optimum (formerly Altice USA), specifically, are facing the same industry wide pressures against a company-specific challenge of pre-existing elevated leverage. 

Company commentary:

Optimum's management framed Q2 2026 as disciplined execution inside a difficult broadband market: broadband losses were held to 40,000 PSUs (aided by a bulk agreement) while Adjusted EBITDA margin expanded 140 basis points to 38.8% and gross margin rose 180 basis points to 71.0%, driven by "lower truck rolls and call volumes, lower sales acquisition costs, and workforce optimization"; management pointed to mobile as its growth lever, citing 40% year-over-year growth in residential mobile revenue and its strongest second-quarter mobile line additions on record, and separately flagged that "demand for Lightpath's AI-grade infrastructure continues to be strong," citing new fiber builds supporting two hyperscale data center campuses.

During the company’s 2Q26 earnings call, management affirmed subscriber losses as arising from: 

• Heightened competitive promotional activity from fiber overbuilders and fixed wireless providers (including Starlink deploying "very aggressive offers")
• Fiber competitor expansion across the company's footprint
  
• Churn particularly intense in the West footprint, where competitive pressure is most severe
  

Lumen's 2026 cash generation has been buoyed by the February 2026 AT&T fiber sale. Having completed the $5.72 billion sale of its Mass Markets Fiber-to-the-Home business to AT&T on February 2, 2026, management is now running the retained network business under a modernization program targeting "$1 billion annualized in cost savings on a run-rate basis exiting 2027,"

Altice France agreed to sell its Mainland France telecom activities to a Bouygues Telecom/Iliad/Orange consortium for €20.35 billion enterprise value (plus a potential earn-out), excluding XpFibre, Intelcia, UltraEdge, Altice Technical Services and overseas operations. An exclusivity agreement (April 16, 2026) preceded a June 6, 2026 Memorandum of Understanding.

Leverage effect: Management's own materials show Adjusted Net Leverage falling from 8.76x (current, reported basis at 2Q26) to a pro forma 5.9x LTM at Q2 2026-end once the transaction (plus the already-closed Intelcia stake sale) is reflected — a reduction of roughly 2.9 turns, funded by using sale proceeds from the €20.35 billion enterprise value to pay down debt, alongside pro forma liquidity of €0.43 billion. Worth flagging: that 5.9x figure is management's own pro forma estimate ahead of closing, not yet a reported number, and is contingent on the deal actually completing on the disclosed terms.

Charter's management was the most direct about near-term pressure, stating plainly that "the competitive environment continued to challenge our Internet customer growth" after losing 172,000 internet customers in the quarter, and attributed the EBITDA decline specifically to "lower revenue and higher transition expenses incurred as we prepare to integrate Cox Communications" — framing this quarter's softer profitability as a temporary cost of the pending Cox acquisition rather than deteriorating core economics, while pointing to mobile line growth and improved video customer retention as offsetting strengths.

Charter's internet subscriber losses (172,000 in the quarter) came alongside management's confirmation of "higher transition expenses incurred as we prepare to integrate Cox Communications."

Section 3: Digital Infrastructure — AI Compute & Power

CoreWeave, Core Scientific and Applied Digital are all monetizing the same AI infrastructure buildout from different points in the stack — GPU cloud compute, converted crypto-to-AI hosting, and purpose-built HPC campuses, respectively — which is why credit profiles diverge sharply despite selling into identical end demand. Equinix sits alongside them as the more traditional counterpart: an investment-grade-adjacent global colocation REIT now capturing the same hyperscaler AI demand through its xScale joint-venture platform, giving the group a lower-levered benchmark against which to read the three AI-native names' leverage.

CoreWeave more than doubled revenue on an LTM basis (+115.3%) as GPU-cloud demand scaled, but leverage remains elevated at 6.82x adjusted / 7.99x total, with coverage thin at 2.86x. FCF is now -$13.66bn LTM — capex has accelerated even faster than the FY25 annual run-rate suggested, as the company continues buying and leasing GPUs well ahead of revenue recognition. Management's own risk disclosures center on AI demand durability — whether "scaling laws" continue to hold, and whether compute demand could shift toward smaller, decentralized deployments — alongside construction risk on the data centers built to house its GPU fleet and the planned NVIDIA Rubin deployment in 2H26.

Core Scientific is mid-conversion from bitcoin mining to AI/HPC hosting, and its 2Q26 LTM leverage metrics remain not meaningful (NM) due to negative FY25 EBITDA.  The notable shift here is revenue: LTM growth has flipped to +26.4%, reversing the FY25 annual decline, as CoreWeave hosting revenue has evidently overtaken the wind-down of legacy mining within the past few quarters. FCF is -$1.2bn LTM. Its CoreWeave hosting agreement is funded almost entirely by the customer, capping Core Scientific's own capex exposure — but new self-funded sites (Hunt County, TX; Muskogee, OK) carry uncapped tariff and construction risk. Management notes it energized more megawatts in 2025 than the rest of its publicly traded peers combined.

Applied Digital's Adjusted Net Leverage reflects healthy debt issuance of circa $5bn in FY2026 (May Year End). Revenue is up 167.5% as new hyperscaler leases ramp, but FCF remains deeply negative at -$2.88bn. Management flagged that its own estimate of hyperscaler AI capex moved from roughly $400bn to nearly $700bn annualized within a single quarter, and separately credited CoreWeave's new investment-grade A3 rating with improving the credit profile of its own Polaris Forge 1 leases.

Applied Digital also highlighted the following operational and pricing outlooks for in the company’s FY26 earnings call.

Operational Highlights

• Polaris Forge 1: 100 MW delivered on time and on budget; additional 75 MW also delivered on schedule.

• Currently constructing 5 multibillion-dollar AI factory campuses for 3 separate hyperscalers.

• All construction projects reported on time and on budget — placing the company in the top 10% of industry performers (industry data shows ~90% of projects over $1bn are delivered late or over budget).

• Expanded from 1 campus to 5 campuses over the past year.

• Visibility to expand to over 5 GW of critical IT load through 2032 using existing campus infrastructure.

ChronoScale (Cloud Business Separation)

• Applied Digital completed the spin-off of its cloud business, now operating as ChronoScale (CHRN), which began trading on NASDAQ in early May 2026.

• Applied Digital retains ~96% ownership of ChronoScale.

• ChronoScale extended customer contracts at higher pricing for its deployed GPU fleet and is pursuing large reserve contracts that could deploy hundreds of megawatts of compute on take-or-pay long-term terms.

• New leadership hires include a CTO with 13+ years at Tesla (VP-level, AI infrastructure) and a CPO with 20+ years in global cloud and AI platforms.

Capital & Financing Strategy

Management outlined a three-part "flywheel" capital strategy:

• Corporate Balance Sheet & Revolver — Initial funding via corporate cash flows and a revolver at SOFR + 225 bps.

• Macquarie JV Equity — A $5 billion JV with Macquarie funds 3/4 of required equity, targeting mid-teens IRR / 1.8x MOIC. Management noted this is preferred over dilutive equity issuance or convertible debt.

• Project Debt — Site-specific construction bonds transitioning to permanent financing (CMBS, ABS, or 144A IG market) as construction risk decreases.

Leverage targets:

• Sub-7x leverage on an 80% loan-to-cost basis for 1.4 GW of contracted capacity.

• Restricted cash balance exceeds $2 billion (covering debt service reserves and letters of credit).

• Management noted private market comparables operate at 10x+ leverage, implying a conservative profile.

Pricing & Margin Outlook

• Contracted lease rates described as "right in the band, if not at the middle towards the higher end" relative to peers.

• Pricing has moved materially higher over the past 6 months; expansion and new campus capacity expected to command premium rates.

• Margin expansion expected through: (1) operational improvements as sites reach Ready-for-Service (RFS) maturity, and (2) economies of scale from multiple buildings within campuses.

• Management noted comparable businesses typically transact at 20–25x NOI multiples.

As a REIT Equinix is the outlier for stability: 3.80x adjusted leverage and 11.08x coverage, the strongest of the four by a wide margin. Management cited record profitability and double-digit MRR growth from xScale leasing, describing demand as "broad-based" and including "emerging customer AI use cases," while flagging power allocation — not capital or chip supply — as the binding constraint; newer IBX designs are now built for twice the power and cooling density of prior generations.  Management raised both full-year 2026 guidance and its 2027–2029 long-term outlook (revenue growth raised to 10–13% annually, adjusted EBITDA margin target raised to 53%+, capex range raised to $5.0–7.0 billion annually), citing stronger-than-expected demand, accelerating bookings/presales, and firm pricing. The commentary attributes this partly to one-time xScale fees flattering the quarter, but the guidance raise suggests management sees the underlying hyperscale/AI demand trend as durable rather than a one-off.

Industry-wide commentary: across CoreWeave, Core Scientific and Applied Digital, a common thread emerges independent of each company's own results. All three now treat power and grid interconnection, not chip supply or capital access, as the binding constraint on growth — Applied Digital cites U.S. Department of Energy estimates that the grid needs roughly 100GW of new capacity by 2030 (about half of it data-center-driven), against a Boston Consulting Group-estimated shortfall exceeding 45GW, while Core Scientific describes competition for power capacity and land as "particularly acute" given limited interconnection capacity at utilities and grid operators. CoreWeave, meanwhile, is the most explicit about demand-side fragility: it frames its entire growth thesis as contingent on AI scaling laws continuing to hold, warning that a plateau in model performance gains, or a shift toward smaller/edge compute, could reduce demand for the large centralized GPU clusters it and its peers are racing to build. All three also carry overlapping hyperscaler counterparty concentration — Core Scientific and Applied Digital both have direct commercial ties to CoreWeave itself — meaning a demand or credit shock at a single hyperscaler would propagate through the smaller infrastructure providers simultaneously.

Disclaimer: This analysis was produced by Cognitive Credit AI and is based on management commentary from 1Q26 earnings calls and Cognitive Credit's curated financial data. Forward-looking statements reflect management views as of the call dates and are subject to change. This is not investment advice. General Mills' latest reported period is 3Q26 (fiscal year-end May), so its leverage and EBITDA data reflects a more recent period than the other companies shown.

Want to try Cognitive Credit AI? Request your demo today.