Is IBM Building the AI Delivery Layer?

IBM (NYSE: IBM) | ~$236 | 52-Week Range: $199.19 – $332.46 | Market Cap: ~$225B

The Big Question

Three frontier model deals in under twelve months. First Anthropic, then Google Cloud in June, and now OpenAI on August 13. Each partnership embeds a different model family into IBM Consulting Advantage, the company’s AI-powered delivery platform. The accumulation is not accidental.

The question institutional investors are debating is not whether IBM can sign model partners. It clearly can. The question is whether aggregating frontier model access inside a consulting delivery platform constitutes a real competitive moat, or whether IBM is assembling a marketing portfolio that sophisticated enterprise buyers will eventually route around by going directly to OpenAI, Anthropic, or Google themselves.

That distinction determines everything: the multiple the stock deserves, the trajectory of the Consulting segment, and whether IBM’s model-agnostic positioning is a durable strategic advantage or a transitional posture that stronger competitors will erode.

Why Wall Street Cares

IBM posted Q2 2026 revenue of $17.16 billion, up just 1.1% year over year, missing the consensus estimate and snapping five consecutive quarters of EPS beats. Operating diluted EPS came in at $2.93 against a $2.97 expectation. Management cited revenue headwinds late in the quarter. The stock is approximately 28% below its June 2 all-time closing high of $329.23.

Within that Q2 result, the segment breakdown tells the real story. Software grew 5% to $7.8 billion, with recurring revenue at 80% of the total and ARR up 8% to $24.6 billion. Red Hat grew 11%. Infrastructure fell 7%, including a 42% collapse in IBM Z mainframe revenue. And Consulting came in at $5.33 billion, essentially flat.

That consulting flatness matters because the AI consulting services market is expanding fast. Institutional investors sitting with IBM in a portfolio today are holding a company whose largest services segment is growing at roughly 1% in a market that industry analysts project will compound at over 26% annually through 2035. The OpenAI deal is IBM’s most explicit attempt to close that gap, which is why it lands in investment committee conversations the morning after it was announced.

The Bull Case

IBM’s argument to institutional investors is that enterprise AI adoption is entering its most consequential phase, and that the bottleneck is no longer model quality. Andy Baldwin, Global Senior Vice President at IBM Consulting, framed it precisely: “The challenge is not access to AI technologies — it’s integrating AI securely and at scale into complex enterprise environments and workflows.”

That framing is correct for a specific and lucrative slice of the market. Global banks, government agencies, and telecommunications carriers operate under regulatory frameworks that require auditability, data residency controls, and governance layers that OpenAI, Anthropic, and Google cannot provide alone. IBM can. Red Hat OpenShift controls roughly 44% of the container orchestration market. IBM Autonomous Security runs multi-agent cybersecurity workflows across enterprise environments. IBM Z mainframes process a significant share of global payment transactions.

Adding GPT-5.6, Codex, and ChatGPT Work to IBM Consulting Advantage closes the remaining model coverage gap, specifically in code generation and frontier reasoning, where OpenAI commands the strongest enterprise preference. Combining that model access with a 130,000-consultant delivery force that is now being trained and certified under OpenAI Partner Network credentials creates something no pure-play AI lab can replicate: a credentialed, governed, regulated-industry deployment machine.

The retraining approach is financially important. IBM is certifying tens of thousands of consultants by retraining existing employees, not by adding headcount. If the incremental revenue follows, the margin improvement could be meaningful. Software ARR was already up 8% through Q2. A re-acceleration toward double digits, driven by watsonx platform attach rates on top of OpenAI-led consulting engagements, is the scenario that justifies analyst targets in the $280–$320 range.

The Bear Case

Analyst targets at Morgan Stanley have been cut to $190 from $293. Goldman Sachs lowered its target to $270 from $335. Citi moved from $255 to $245. The directional pattern in recent revisions is uniformly downward, and the reasoning goes beyond one weak quarter.

The structural bear argument is that IBM’s model-agnostic positioning, while intellectually coherent, is not a strategy that compounds. Every frontier model IBM adds to its delivery platform is also available to Accenture, Infosys, TCS, Deloitte, and Capgemini. Accenture surpasses IBM Consulting in scale and operates with greater vendor neutrality. Infosys is already collaborating with Google and Microsoft on AI-native enterprise solutions and building industry-specific consulting capabilities through acquisitions. None of IBM’s model partnerships are exclusive.

The enterprise disintermediation risk is real. OpenAI has signed its own direct partnerships with Infosys and Tata Consultancy Services as it builds out its enterprise deployment network. If large enterprises decide they can work directly with OpenAI through those channels, or simply negotiate enterprise agreements with the model providers themselves, IBM’s consulting delivery layer becomes a cost center rather than a value-add. The IBM Z mainframe, which anchors IBM’s stickiness with global financial institutions, is currently in a cycle trough with revenue down 42% in Q2. That anchor weakens as cloud migration accelerates.

The break condition is visible: if Q3 consulting signings growth decelerates below 3% and the revenue-to-signings conversion gap widens rather than narrows, the thesis collapses. Consulting signings rose 5% to $5.034 billion in Q2, which means the order book is building. But the time between signing and recognized revenue in large consulting engagements is long, and IBM’s guidance has already been lowered once this year.

The Evidence

Three data points anchor the current debate.

First, watsonx cumulative AI book of business exceeded $12 billion inception-to-date through early 2026, a figure that surprised analysts who had written off IBM’s AI relevance. That number confirms enterprises are past experimentation. It does not confirm that IBM is capturing its share of the production-scale deployment cycle that follows.

Second, within IBM’s Q2 software results, the high-growth portfolio — Red Hat, watsonx, HashiCorp, Confluent — is delivering consistently. Red Hat grew 11%. Data grew 19%. Distributed Infrastructure, including Power and Storage, built an order backlog of nearly $500 million in Q2. The software engine is working. The problem is that software alone, at current growth rates, cannot offset consulting stagnation at IBM’s scale.

Third, IBM’s own internal AI deployment data offers an early signal. The company’s Bob AI coding tool, which draws on a mixture of generative models, reached adoption across more than 80,000 IBM employees during Q2. That internal scale is both a proof of concept and a sales reference. IBM is certifying its own consultants on the same stack it is selling, which matters when regulated enterprise buyers want demonstrated operational experience rather than vendor promises.

The Mavens’ View

The most sophisticated read of IBM’s position right now is not bullish or bearish in the conventional sense. It is conditional.

Portfolio managers who remain constructive on IBM are not betting on a consulting turnaround in the traditional sense. They are betting on a specific structural shift: that regulated-industry enterprises, particularly in financial services and government, will consolidate their AI deployment spend with a small number of trusted integrators over the next 18 to 24 months, and that IBM’s combination of governance tooling, hybrid cloud infrastructure, mainframe credibility, and now multi-model delivery will make it one of the two or three firms that win that consolidation.

The cautious view, represented by the Morgan Stanley downgrade and the Citi target reduction, holds that IBM’s consulting flatness is a structural signal, not a cyclical one. If IBM’s 130,000-person consulting force cannot grow revenue in a market expanding this quickly, adding another model partner does not solve the underlying problem. The concern is not that IBM lacks access to frontier models. The concern is execution: certification speed, sales force enablement, and the ability to win competitive decisions where Accenture and Deloitte are also showing up with their own frontier model credentials.

Where the professional investor community appears to agree: the Q3 earnings call in October is the most important data event IBM has in 2026. Consulting revenue trajectory, AI book composition, and any update on OpenAI Practice certification volumes will determine whether the current price represents a recovery entry or a value trap.

What Investors Are Missing

The coverage on IBM’s OpenAI deal has focused almost entirely on consulting revenue. That framing misses the more important second-order dynamic.

IBM is not just selling consulting hours with better AI tools. It is building certification infrastructure. OpenAI committed to certifying 300,000 consultants through its Partner Network by the end of 2026, with $150 million committed to the partner ecosystem. IBM, as an Elite partner, sits at the top of that certification hierarchy. That means IBM’s consultants carry credentials that OpenAI’s own go-to-market motion will increasingly reference when regulated-industry customers ask which partners can deploy responsibly at scale.

This creates a dynamic that the revenue-per-quarter framing misses entirely. IBM is becoming a primary node in OpenAI’s enterprise distribution network for the verticals OpenAI cannot reach alone: government procurement environments, Basel-regulated financial institutions, and national telecommunications carriers with data sovereignty requirements. If OpenAI’s enterprise revenue scales at the rate its partner investment suggests, IBM’s Elite status becomes a privileged position in that distribution chain, and the revenue it captures will look more like platform licensing than consulting hours.

Nobody covering IBM is modeling that. The bear case assumes IBM consultants are commodity. The real possibility is that IBM-certified, OpenAI-credentialed, watsonx-governed delivery is a category that commands premium pricing in markets where governance is non-negotiable, and that the consulting revenue line understates what IBM is actually building.

Stocks to Watch

IBM (NYSE: IBM) — The central debate. At roughly $236, the stock prices a company whose consulting recovery is still theoretical. The upside case requires Q3 consulting reacceleration and evidence that signings are converting. The downside case requires only that Q3 looks like Q2. The October earnings call is the decisive event. Analyst consensus sits at a median $240 with a range from $174 to $350, reflecting genuine uncertainty rather than analytical precision.

Accenture (NYSE: ACN) — IBM’s most direct competitor for regulated-industry AI consulting mandates. Accenture operates with greater vendor neutrality and larger scale, but does not carry IBM’s proprietary infrastructure stack. If IBM’s Elite OpenAI status translates into competitive wins in financial services and government, those wins come at Accenture’s expense. Watch Accenture’s AI consulting revenue growth and any commentary on competitive dynamics in regulated verticals at its next earnings update.

Infosys (NYSE: INFY) — The overlooked competitor in this dynamic. Infosys holds its own OpenAI partnership and is building industry-specific consulting capabilities through targeted acquisitions. Its revenues grew 3.2% year over year in its most recent quarter. If IBM’s model-agnostic positioning becomes table stakes rather than a differentiator, Infosys is positioned to compete directly in the same enterprise segments at lower cost. It is the highest-risk name for IBM’s consulting thesis if the certification moat does not hold.

Red Hat (IBM subsidiary, traded via IBM) — The most underappreciated dimension of the IBM platform story. Red Hat OpenShift controls approximately 44% of the container orchestration market. Every enterprise AI deployment that runs on a hybrid cloud environment is a potential OpenShift workload. As OpenAI-powered consulting engagements generate production deployments, Red Hat’s infrastructure becomes the hosting layer. Red Hat grew 11% in Q2. If AI deployment volumes accelerate through IBM Consulting, Red Hat’s growth rate could move meaningfully higher, and it would show up in IBM’s software segment ARR before it shows up in consulting revenue.

Palo Alto Networks (NASDAQ: PANW) — The cybersecurity angle on this deal is underweighted. IBM’s OpenAI partnership includes a specific integration with IBM Autonomous Security through the OpenAI Daybreak Cyber Partner Program. As enterprises deploy AI at scale, the attack surface expands, and governed security workflows become a procurement requirement. IBM is positioning itself in that spend. Palo Alto Networks competes for the same enterprise security budget. If IBM’s multi-agent security platform gains traction through the consulting channel, Palo Alto faces a credentialed integrator competitor in accounts where it previously had limited overlap with IBM.