Key Takeaways
- OpenAI President Greg Brockman said on November 5, 2026 that enterprise customers now generate the majority of OpenAI’s revenue, and that the company is targeting a 2027 IPO.
- Global enterprise AI spending reached roughly $177 billion in 2026, a 5.4x year-over-year increase, according to Menlo Ventures‘ fourth annual “State of Enterprise AI” report published November 3, 2026.
- Only about 22% of that spend came from traditional software budgets, indicating AI is largely net-new expenditure rather than replacing existing SaaS contracts.
- Incumbents are buying AI capability at premium multiples: SAP‘s acquisition of 18-month-old Prior Labs, which had raised €9 million in seed funding, closed July 17, 2026 at a reported value above €1 billion.
- Concentration and cost risks persist: roughly 80% of OpenAI and Anthropic revenue comes from 1% of customers, and one in three enterprises imposed an emergency AI spending freeze in 2026.
Enterprise Becomes OpenAI’s Majority Business
OpenAI President Greg Brockman disclosed on November 5, 2026 that enterprise customers now account for the majority of the company’s revenue and that OpenAI is targeting a 2027 initial public offering, a pivot confirmed the same week that Menlo Ventures reported global enterprise AI spending reached roughly $177 billion in 2026, a 5.4x increase year over year.
The two data points arrived within 48 hours of each other and describe the same shift from opposite ends: the buyer side is spending at an unprecedented rate, and the leading model developer is now financially dependent on that spending.
Key Facts
Menlo Ventures published its fourth annual “State of Enterprise AI” report on November 3, 2026, putting 2026 enterprise AI spend at $176.8–177 billion, up 5.4x year over year. Approximately 22% of that spending came from traditional software budgets, per SaaStr’s November 4 analysis of the Menlo data; the remainder represents net-new spend.
Anthropic captured roughly 55% of enterprise model API spending, about six times Google’s share, according to Menlo data. The average firm ran 2.5x over its AI budget in 2026, and one in three enterprises instituted an emergency AI spending freeze, per AIM Media House. PYMNTS reported on September 4 that OpenAI and Anthropic each derive about 80% of revenue from 1% of customers.
Brockman, speaking November 5, described the first half of 2026 as “rough” before a coding-AI takeoff revived growth, according to The Wall Street Journal. Bloomberg reports OpenAI still expects $70 billion in annualized revenue by the end of 2026, after annualized revenue came in $20 billion below prior investor estimates in early October.
Incumbents Buy Their Way Into AI
The SAP–Prior Labs transaction has become the archetype of the current acquisition wave. SAP agreed on May 4, 2026 to acquire Prior Labs, a German tabular-foundation-model startup, alongside data-lakehouse vendor Dremio; the Prior Labs deal completed July 17, 2026.
Deal value was reported at roughly €1 billion or more ($1.16 billion), with SAP committing more than €1 billion to scale the lab, according to TechCrunch and Tech.eu. Prior Labs was 18 months old and had raised just €9 million in a February 2025 seed round, a markup exceeding 100x.
The strategic rationale is instructive: Prior Labs builds foundation models for tabular data such as ERP tables, not chatbots. SAP CTO Philipp Herzig framed structured data as “the greatest untapped opportunity in enterprise AI” in comments reported by EME Outlook on July 20, 2026. For SAP, the purchase was less about acquiring a product than about owning the model layer beneath its own systems of record.
The broader wave is moving quickly. Salesforce acquired Cresta for a reported ~$2.7 billion on November 3, roughly 1.7x its 2022 valuation, per The Information. Workday acquired C1.ai in October. CrowdStrike reportedly bought AI security startups for $495 million. 36Kr counted 195 verified AI M&A deals globally, with OpenAI the most active acquirer.
The macro context explains the urgency: US venture deal value hit a record $515.8 billion while exits lag, making acquisitions the exit of choice, according to SiliconANGLE. For incumbents, buying an 18-month-old startup at a billion-dollar valuation is faster than building a model team from scratch.
The Revenue Engine and Its Concentration Risk
Enterprise is now OpenAI’s majority revenue source, per Brockman, but the customer base is thin: roughly 80% of OpenAI and Anthropic revenue comes from 1% of customers, according to PYMNTS. That concentration cuts both ways: a handful of large accounts can swing quarterly results, and each account carries negotiating leverage that smaller buyers do not.
Cost overruns compound the risk. The average firm ran 2.5x over its AI budget in 2026, and one in three enterprises froze AI spending outright, per AIM Media House, evidence that CFOs did not plan for this spend and are now applying brakes mid-cycle.
Price competition is intensifying. OpenAI is winning back enterprise customers from Anthropic on price, The Wall Street Journal reported on October 7, and Replit switched its default model provider from Anthropic to OpenAI on October 17, per TechCrunch. Anthropic, meanwhile, filed an S-1 in late October 2026 for a potentially record-breaking IPO, with revenue growth slowing and a reported net loss exceeding $80 billion in H1 2026 on an accounting basis, according to Reuters and The Guardian.
Lock-in concerns are rising alongside the acquisitions. Gartner‘s May 12 First Take flagged that SAP’s Dremio and Prior Labs deals “raise third-party vendor restriction concerns” for buyers wary of single-vendor dependence. As deployment services become the differentiator, AWS and Anthropic are racing to deploy “an army of specialized engineers,” CNBC reported on October 15, a signal that the moat is shifting from model quality to implementation capacity.
Industry Reaction and Expert Perspectives
Brockman framed the shift plainly: enterprise is now the majority of sales and the company is “ready to IPO in 2027,” he said, per Reuters on November 5, 2026.
Herzig of SAP positioned his company’s purchase around data rather than chat, calling structured data “the greatest untapped opportunity in enterprise AI,” per EME Outlook on July 20, 2026.
Menlo Ventures’ data, as cited by SaaStr on November 4, 2026, carries a structural implication: AI spend reached $177 billion, but only 22% came out of software budgets, meaning AI is additive to IT spending, not a substitution for it.
Gartner’s May 12, 2026 note added a cautionary thread, warning that vendor-restriction concerns surround SAP’s acquisitions as buyers weigh dependence on a single platform.
One sourcing caveat: Menlo Ventures is a venture firm with portfolio interests in the sector, and SaaStr and FourWeekMBA are secondary aggregators of its data. The Menlo report itself should be cited directly for the spend figures.
The Bottom Line
The enterprise AI market is accelerating late in its cycle: incumbents are paying premiums for capability they cannot build fast enough, and the two leading model developers now depend on enterprise revenue concentrated in a tiny customer base. Watch three signals: whether more 18-month, billion-dollar acquisitions follow Prior Labs; whether the OpenAI–Anthropic price war compresses model margins and shifts differentiation toward deployment services; and whether OpenAI’s 2027 IPO target holds after a turbulent 2026. For buyers, the practical guidance is to budget for 2.5x overruns and negotiate multi-model portability before signing.
