Tech M&A Watch: SAP's Dremio Bet, Nscale-Anyscale, and the Infrastructure Land Grab Beneath AI
Owning the Layer Beneath the Model — the Infrastructure Land Grab
SAP buys the data foundation agentic AI needs to work, Nscale and IREN acquire their way up the compute stack, Teledyne pays an 88% premium for next-generation imaging IP, and Palo Alto Networks folds identity security into its AI-era platform. Why today's buyers are betting the durable advantage sits one layer below the model.
Global dealmaking has strengthened through 2026, with the AI buildout pulling merger activity toward semiconductors, energy, digital infrastructure, and technology platforms, even as overall deal volume stays selective. What matters this cycle isn't the count of transactions — it's what they reveal. Company after company is acquiring capabilities it cannot build fast enough on its own, and almost none of today's biggest deals are about the AI model itself. They're about the layer directly beneath it: the data, the compute orchestration, the network, and the supply chain that determine whether AI capability is actually usable.
Seven Transactions That Define the Cycle
Teledyne acquires Varex Imaging
$1.1B · All-cash · ~88% premiumBuyer → Teledyne. Target → Varex Imaging, at $18.90 per share. Rationale → Broadens Teledyne's healthcare and sensing portfolio across medical, dental, and life-science imaging — a supply-chain and technology-ownership move, acquiring next-generation CT and imaging-detector capability outright rather than licensing it. Competitive impact → Gives Teledyne direct ownership of the next-generation CT detector supply chain; expected to close in early 2027.
Nscale acquires Anyscale
$1.65B · Largest disclosed deal in this cycleBuyer → Nscale, a neocloud provider. Target → Anyscale, an AI compute-software startup. Rationale → Nscale is seeking to own more of the AI compute stack from hardware up through the orchestration layer, rather than differentiate on raw capacity alone. Competitive impact → The clearest signal yet that infrastructure providers are moving up the stack to differentiate through software, not just GPU count.
SAP completes acquisition of Dremio
Terms undisclosed · Closed 2026Buyer → SAP. Target → Dremio, an Apache Iceberg-native data lakehouse built to power agentic AI. Rationale → SAP's stated goal is to unify SAP and non-SAP data so AI agents operate on governed, AI-ready context instead of fragmented data. SAP's CTO framed it directly: enterprise AI stalls not because models aren't good enough, but because the data isn't ready for AI agents. Competitive impact → The defining transaction of this cycle — converts SAP's massive installed customer base into a base for defensible, AI-driven recurring revenue.
IREN completes acquisition of Mirantis
12.6M shares + ~$40M cash · Equity-heavy structureBuyer → IREN. Target → Mirantis, a cloud-software provider serving more than 1,500 enterprise customers with deep Kubernetes (k0rdent) expertise. Rationale → Strengthens the software layer of IREN's vertically integrated AI cloud platform. Competitive impact → The equity-heavy structure aligns Mirantis to IREN's AI-cloud upside rather than simply cashing sellers out — a signal of confidence in IREN's own valuation trajectory.
OpenAI acquires NextSlide
Terms undisclosed · Closed earlier in 2026Buyer → OpenAI. Target → NextSlide, whose AI product turns prompts, notes, and research into polished presentations; the team has joined the ChatGPT organization. Rationale → A revenue- and product-surface-driven tuck-in — the latest in a string of OpenAI capability acquisitions that also includes open-source tooling maker Astral. Competitive impact → Strengthens ChatGPT's footprint in the workplace-productivity category with a ready-made, monetizable product rather than a research capability.
Palo Alto Networks completes acquisition of CyberArk
Identity consolidation · Closed 2026Buyer → Palo Alto Networks. Target → CyberArk — privileged access, machine identity, and secrets management. Rationale → Identity has become the control plane for agentic AI, where human, machine, and autonomous identities must all be secured; this converts a discrete security product into a platform bet. Competitive impact → Positions Palo Alto to capture a larger share of enterprise security budgets by owning the identity surface every organization must secure.
Lumen acquires Alkira; Belden agrees to acquire RUCKUS Networks
Two separate deals · Networking consolidationBuyer → Lumen (Alkira); Belden (RUCKUS Networks). Rationale → Lumen is accelerating an AI-era unified digital platform; Belden is moving toward full-stack IT/OT networking. Both consolidate networking software at a moment when AI traffic is redefining network demands. Competitive impact → Signals that networking — often treated as commodity plumbing — is being actively fought over as AI workloads change what the network needs to do.
Revenue-Driven or Capability-Driven? Increasingly Both
The defining pattern of 2026 is that AI acquisition strategy has split into two distinct motives. The first is revenue-driven: companies buying AI products that can be monetized immediately. OpenAI's NextSlide acquisition fits here — a consumer and workplace product that strengthens ChatGPT's surface area and gives the company a ready-made position in the productivity category.
SAP's purchase of Dremio is the archetype. SAP is not buying an AI model — it's buying the data foundation that makes agentic AI viable, because no amount of internal R&D can assemble governed, fragmented enterprise data from scratch faster than a proven lakehouse can. IREN's acquisition of Mirantis applies the same logic to infrastructure software: orchestration, monitoring, and customer-support layers that turn raw compute into a usable cloud.
Across these deals, the answer to whether companies are buying AI for revenue or for strategic capability is increasingly both — with capability acquisition dominating at the frontier. MongoDB's earlier $220 million purchase of Voyage AI, integrating embedding and reranking models into its database to curb AI hallucinations, is a capability play dressed as a product extension: it reads like a feature launch but functions as a defensive acquisition of a specific technical bottleneck.
The Most Active Arena in Technology Right Now
The clearest strategic consolidation of this cycle is in infrastructure. The Teledyne-Varex transaction shows continued consolidation around imaging and sensing technologies, giving Teledyne ownership of the next-generation CT detector supply chain — a supply-security motive: in a world of constrained components, owning the detector source is a competitive advantage in itself, independent of any AI narrative.
The infrastructure story runs broader than semiconductors, though. IREN-Mirantis and Nscale-Anyscale both show compute providers building vertically, integrating data centers, compute, and software into a single stack rather than competing on hardware access alone. d-Matrix's acquisition of Wallaroo.ai extends the same pattern into inference orchestration, speeding deployment of heterogeneous AI inference workloads.
AI demand is accelerating consolidation across compute, networking, storage, software, and power simultaneously — not sequentially. That's what makes the infrastructure layer the most active M&A arena in technology right now: every sub-layer is being fought over at once, from Teledyne's detector supply chain to Nscale's orchestration software to Lumen and Belden's networking consolidation.
Customers → Data → AI Capabilities → Distribution → Recurring Revenue
Enterprise software buyers are consolidating around a clear value chain, and SAP's Dremio acquisition is the defining transaction of the pattern. SAP already owns the largest base of enterprise customers and their core operational data. What it lacked was a unified, open data foundation to feed agentic AI. Dremio supplies that data layer; the payoff is defensible, recurring AI-driven revenue on top of SAP's existing distribution — consolidation designed explicitly to convert installed customers into AI customers.
Cybersecurity follows the same logic from a different angle. Palo Alto's CyberArk deal is less about acquiring new technology than about owning the identity surface every enterprise must secure, converting a discrete security product into a platform that captures a larger share of a big, recurring security budget. In both cases, the acquirer already had the customers and the distribution — what it bought was the specific capability layer needed to keep monetizing that base as the product category shifts toward AI.
Why Acquisition Won in Each of Today's Biggest Deals
Nscale-Anyscale: time-to-market. The AI buildout moves faster than any internal engineering timeline allows — Nscale could not build Anyscale's orchestration layer in the time the AI-cloud race permits, so it bought it.
SAP-Dremio: proprietary technology and IP. Dremio's stewardship of Apache Iceberg, Arrow, and Polaris is hard-won open-source leadership that SAP could not replicate quickly through internal development alone.
Teledyne-Varex: proprietary technology and IP. Varex's next-generation CT detector technology is protected, specialized IP built over years — buying it outright was faster and lower-risk than a competing R&D program.
OpenAI-NextSlide and OpenAI-Astral: talent. Both are explicitly talent-and-product acquisitions — acquiring teams already operating at the frontier of their category rather than hiring and training a comparable team from scratch.
SAP-Dremio (again, on data): the decisive asset in enterprise AI is governed data. Buying Dremio buys access to an entire open data ecosystem and semantic layer that internal development would take years to reproduce.
Across the board — competitive pressure: when hyperscalers and neoclouds are consolidating the compute stack, inactivity is a losing strategy. Infrastructure players are buying specifically to keep pace with rivals who have already acquired the controlling layers.
Who Is Reshaping Their Stack Through M&A
The most aggressive acquirers this cycle cluster into two groups. The first is hyperscalers and compute providers: Nvidia has continued its pattern of acquiring developer and AI-cloud tooling (Brev.dev), while compute providers like IREN, Nscale, and d-Matrix are buying software specifically to move up the stack from raw capacity toward orchestration and deployment. OpenAI has become a serial acquirer in its own right, completing roughly six acquisitions in 2026.
The second group is enterprise software and security platforms: SAP (Dremio), Palo Alto Networks (CyberArk), and Lumen (Alkira) are each consolidating their respective layers — data, identity, and networking — to defend and expand their existing positions rather than enter new categories. These buyers share one trait: they are reshaping their technology stacks through M&A rather than building, because the pace and stakes of the AI transition make purchasing faster and safer than invention.
Strategic Premiums, Paid Increasingly in Equity
The economics of today's deals span the spectrum. Teledyne-Varex is an all-cash strategic acquisition at a premium near 88% — the kind of premium a buyer pays specifically for supply security and next-generation technology, not growth metrics. Nscale-Anyscale is the largest disclosed deal in this slate at $1.65 billion, a strategic premium for a full AI compute stack rather than a single component.
IREN-Mirantis combines stock (12.6 million shares) with cash and other consideration — an equity-heavy structure that aligns the seller to the buyer's AI-cloud upside rather than simply cashing them out. SAP-Dremio and OpenAI-NextSlide are both terms-undisclosed, reflecting private-market negotiations where valuation is secondary to strategic fit. Across the cycle, strategic acquisitions dominate, and they're increasingly paid for with equity in high-multiple AI-cloud companies rather than pure cash — signaling that acquirers are confident enough in the value of their own currency to use it as the deal vehicle. Financially motivated consolidation remains present but secondary to strategic, capability-driven deals.
Three M&A Signals
The strongest structural signals emerging from today's transactions:
AI Capability Acquisition
Companies are buying specialized AI and data capabilities rather than building everything internally. SAP-Dremio and IREN-Mirantis demonstrate that the scarcest assets — governed data and orchestration software — are being acquired, not invented.
Infrastructure Consolidation
AI demand is accelerating consolidation across compute, networking, storage, software, and power. Nscale-Anyscale, IREN-Mirantis, d-Matrix-Wallaroo, Lumen-Alkira, and Belden-RUCKUS all point to compute and networking consolidating into vertically integrated stacks.
Data as the Moat
The most strategically valuable acquisitions aren't for models but for the data layer that makes models useful. Major software platforms are buying the data infrastructure that converts existing customer relationships into defensible AI revenue.
"What does today's M&A activity reveal about where technology companies believe future competitive advantage will come from?"
The infrastructure and data layer beneath the AI model — not the model itself. The model layer is crowded and commoditizing; the defensible value is accumulating one level down. That's why the most aggressive acquisition interest is concentrated in data infrastructure (SAP-Dremio), compute orchestration (Nscale-Anyscale, IREN-Mirantis), networking software (Lumen-Alkira, Belden-RUCKUS), and the supply chain underpinning all of it (Teledyne-Varex, d-Matrix-Wallaroo). The strategic logic is uniform across every deal in this slate: own the layer through which AI value actually flows. Whether through governed data, orchestration software, the network, or the detector supply chain, today's buyers are placing their bets on the enabling infrastructure of the next phase of the technology industry. Companies that can control the data, compute, and connectivity on which AI runs will command the durable advantage of this cycle — and the market is paying strategic premiums, increasingly in equity rather than cash, to secure it now.
Sources
- BCG — Global M&A Rebound Fueled by AI in 2026
- Reuters — Teledyne to acquire Varex Imaging in $1.1 billion deal
- TechCrunch — Coverage of Nscale's acquisition of Anyscale
- SAP News — SAP completes acquisition of Dremio
- GlobeNewswire — IREN completes acquisition of Mirantis
- Bloomberg — Coverage of OpenAI's NextSlide and Astral acquisitions
- Palo Alto Networks — Palo Alto Networks completes acquisition of CyberArk
- Nscale — Company statements on the Anyscale transaction
- MongoDB — MongoDB acquisition of Voyage AI
Reviewed by Erwin Castro
on
Tuesday, August 11, 2026
Rating: