A definitive analysis of fundraises, secondary transactions, and M&A across AI and software — covering $300B+ in deals from 2023 to present. Prepared to investment-grade standards with sourced data from Bloomberg, TechCrunch, PitchBook, SEC filings, and company disclosures.
In April 2026, Anthropic hit $30B annualized revenue — surpassing OpenAI's $25B. From $1B to $30B in 15 months. No enterprise software company has ever grown this fast. 80% enterprise revenue mix vs. OpenAI's consumer-heavy composition.
OpenAI's March 2026 round raised $122B from SoftBank, Amazon, and Nvidia at an $852B valuation — the largest private funding round in history. Much of it in GPU credits and milestone-contingent pledges, not pure cash.
Big Tech spent $23B+ on licensing-plus-talent deals: Google's Character.AI ($2.7B) and Windsurf ($2.4B), Microsoft's Inflection ($650M), Amazon's Adept ($330M), Meta's Scale AI ($14.3B). A new M&A playbook to bypass antitrust.
SpaceX ($1.5T target), OpenAI ($1T), and Anthropic ($380B+) may IPO in 2026. Combined market cap of ~$2.9T would require absorbing more capital than the entire US IPO market raised from 2016-2025 combined.
| Company | Category | Primary Val ($B) | Secondary Val ($B) | Prem/(Disc) | Demand | IPO Timeline |
|---|---|---|---|---|---|---|
| Anthropic | Foundation Model | $380B | $500B+ | +32% | Insatiable — no sellers | Oct 2026 |
| SpaceX | Space/Defense | $800B | $850B | +6% | Very High | Jun 2026 (roadshow) |
| OpenAI | Foundation Model | $852B | ~$750B | -12% | Dropping — $600M+ blocks unsold | Q4 2026 / Q1 2027 |
| Databricks | AI Infrastructure | $134B | $145B | +8% | High | 2026-2027 |
| Cursor | AI Applications | $29.3B | ~$50B | +71% | Moderate (limited float) | 2027+ |
| Anduril | Defense Tech | $30.5B | ~$66B | +116% | Limited supply | 2026-2027 |
| Perplexity | AI Applications | $9B | $11B | +22% | Moderate | 2026-2027 |
| xAI | Foundation Model | $75B | ~$60B | -20% | Moderate | IPO planned 2026 |
| Stripe | Fintech/SaaS | $91.5B | ~$85B | -7% | Moderate | 2026-2027 |
| Harvey | AI Applications | $8B | ~$10B | +25% | Limited | 2027+ |
| Clay | Sales Tech/SaaS | $1.3B | ~$2B | +54% | Limited | 2027+ |
| Figure AI | Robotics/AI | $39B | ~$45B | +15% | Limited | 2027+ |
| Rippling | HR Tech/SaaS | $13.5B | ~$15B | +11% | Moderate | 2026+ |
| CoreWeave | AI Infrastructure | $35B (IPO) | ~$25B | -29% | Moderate | IPO'd Mar 2025 |
The most striking signal: Anthropic trades at a 32% premium to its last primary round while OpenAI trades at a 12% discount. Per Rainmaker Securities, Anthropic is the "most difficult stock to source" with virtually no sellers — while ~$600M of OpenAI shares remain unsold. This divergence reflects the market reassessing which company has the stronger enterprise position, not just the larger user base.
Google, Microsoft, Amazon, and Meta paid an average of ~$100M per engineer in these quasi-acquisitions. The per-head cost reflects the extreme scarcity of world-class AI researchers and the strategic urgency to consolidate talent before competitors do.
OpenAI: 17 acquisitions across 7 categories — "breadth procurement" to support a consumer + enterprise IPO narrative. Anthropic: 4 precision acquisitions, each targeting a specific product layer. OpenAI buys narrative; Anthropic buys barriers.
Regulators have opened probes into Microsoft-Inflection, Amazon-Adept, and scrutinized all major quasi-acquisitions. The fundamental question: are these structured to evade antitrust? The current administration's posture suggests enforcement remains uncertain.
| Acquirer | Target | EV ($B) | Date | Target ARR | EV/Rev | Type | Strategic Rationale |
|---|---|---|---|---|---|---|---|
| Synopsys | Ansys | $35B | Jan 2024 | $2.3B | 15.2x | Full Acq | Silicon-to-systems; AI chip design enablement |
| Wiz | $32B | Mar 2025 | $1B | 32x | Full Acq (Cash) | Multi-cloud security; Google's largest ever | |
| Cisco | Splunk | $28B | Mar 2024 | $3.8B | 7.4x | Full Acq (Cash) | Security + observability; recurring revenue |
| Nvidia | Groq (rumored) | $20B | 2025 | N/A | N/A | Potential | Consolidate inference chip market |
| HPE | Juniper Networks | $14B | Jan 2024 | $5.6B | 2.5x | Full Acq | AI-driven networking; compete with Cisco |
| Meta | Scale AI (49%) | $14.3B | Jun 2025 | $1.4B | ~10x | Strategic Invest | Data labeling for Llama; CEO → Meta |
| IBM | Confluent | $11B | Dec 2025 | $900M | 12x | Full Acq | Real-time data for AI workloads |
| CoreWeave | Core Scientific | $9B | 2025 | N/A | N/A | Full Acq | Power infra for AI; vertical integration |
| Blackstone/Vista | Smartsheet | $8.4B | 2024 | $1B | 8.4x | Take Private | PE play; enterprise SaaS optimization |
| Salesforce | Informatica | $8B | May 2025 | $1.6B | 5x | Full Acq | Agent-ready data; governance for AI |
| OpenAI | io (Jony Ive) | $6.5B | May 2025 | N/A | N/A | Full Acq (Stock) | Consumer AI hardware; post-screen devices |
| IBM | HashiCorp | $6.4B | Apr 2024 | $600M | 10.7x | Full Acq (Cash) | Hybrid cloud; Terraform + Vault + Red Hat |
| AMD | ZT Systems | $4.9B | 2024 | N/A | N/A | Full Acq | Full-stack AI systems; hyperscale |
| ServiceNow | Moveworks | $2.85B | Nov 2024 | N/A | N/A | Full Acq | AI agents for IT service management |
| Character.AI | $2.7B | Aug 2024 | N/A | N/A | License + Hire | Noam Shazeer return; chatbot for Gemini | |
| Windsurf | $2.4B | Jul 2025 | $82M | ~29x | License + Hire | Coding AI talent for DeepMind | |
| Microsoft | Inflection AI | $650M | Mar 2024 | N/A | N/A | License + Hire | Suleyman → MSFT AI CEO; safety talent |
| Amazon | Adept AI | $330M | Jun 2024 | N/A | N/A | License + Hire | AI agent tech for AWS; 66/100 staff hired |
| NXP | Kinara.ai | $307M | 2025 | N/A | N/A | Full Acq | Edge AI processors; India deeptech |
2025's M&A was fundamentally different from 2023-2024 venture funding. VC bet on frontier models and talent. Strategic M&A bet on infrastructure — the unglamorous but essential layers that determine which enterprises deploy agentic AI at scale. Google bought Wiz ($32B) for multi-cloud visibility. IBM bought HashiCorp ($6.4B) and Confluent ($11B) for data pipelines. Salesforce bought Informatica ($8B) for data governance. The pattern is unmistakable: infrastructure determines winners, not model quality alone.
Six major quasi-acquisitions in 18 months: Inflection→Microsoft, Adept→Amazon, Covariant→Amazon, Character.AI→Google, Windsurf→Google, Scale AI→Meta. Combined value: $23B+. Average per-head cost: ~$100M for senior AI researchers. FTC/DOJ probes have not stopped the trend.
CoreWeave acquired Core Scientific ($9B) to own power infrastructure. AMD acquired ZT Systems ($4.9B) to own systems design. As AI scaling becomes constrained by power, networking, and cooling — not processor count — controlling the full vertical stack is the new competitive moat.
Blackstone/Vista took Smartsheet private ($8.4B). PE firms are targeting mature SaaS companies with $1B+ ARR, strong retention, and room for AI-driven operational optimization. The playbook: buy at 7-10x revenue, implement AI, re-list at 15-20x.
With SpaceX, OpenAI, and Anthropic all planning 2026 IPOs requiring $240B+ combined absorption, the secondary market is reshaping M&A calculus. Companies that can't IPO will become acquisition targets. The $2.9T combined market cap of the "Big Three" IPOs will set new public market benchmarks for every AI company.