This week, enterprise AI adoption takes center stage — with long-awaited (and surprisingly confusing) data points from Microsoft.OpenAI’s upcoming funding round offers fresh signals on where the company’s spending might be headed next.
And apparently, we may need to start tracking planetary alignments as part of investment research — because they now seem to influence IPO and M&A timelines.
Enterprise AI: top-down spend vs bottom-up adoption
Enterprise AI adoption is throwing off some confusing signals right now.
This week, Microsoft disclosed for the first time that Office 365 Copilot has 15 million paying subscribers. The figure includes only seats billed separately, starting at $30 per user per month. On that basis, Microsoft is well ahead of Google, which reportedly has around 8 million Gemini Enterprise subscribers, priced from $21 per seat.
On revenue alone, Copilot likely outpaces Anthropic’s chatbot subscription business. At face value, this looks like a decisive win for top-down enterprise AI adoption.
But usage tells a more complicated story.
Anthropic itself is a good example. Despite building some of the most advanced AI tools in the market, the company still relies heavily on legacy enterprise software. A recent job posting for head of IT regulatory compliance lists experience with tools like Workday, Salesforce, and NetSuite. Anthropic has also publicly explained why it chose to partner with Intercom for customer service instead of building an internal AI agent.
Even culturally, the tension is visible. A Claude design lead recently wrote that the “worst part” of company growth is the gradual shift to enterprise-grade internal tools “no one likes using.” The post was later deleted, but the message resonated: incumbents persist, even when better tools exist.
The issue is this: enterprise spend does not equal enterprise usage.
According to The Information, Microsoft leadership has privately raised concerns that Office 365 Copilot is not yet delivering meaningful automation. Some large customers are paying for Copilot seats but using the product sparingly — or not at all.
At the same time, OpenAI — long the driver of bottom-up adoption inside enterprises — is exploring both strategies. The company is preparing new offerings aimed directly at large organizations, designed to embed AI across workflows like customer support and legacy code refactoring.
OpenAI’s round is coming together — fast
OpenAI’s next fundraising round is accelerating.
On Wednesday, reports surfaced that OpenAI is in talks to raise up to $60 billion from a group that includes Nvidia, Microsoft, and Amazon. A day later, The Wall Street Journal reported that Amazon alone could invest as much as $50 billion.
That would be a meaningful commitment for Amazon. As of September 30, the company held $94 billion in cash and marketable securities — meaning this would represent a material deployment of balance-sheet capital.
Amazon will almost certainly want more than financial exposure in return.
OpenAI has already committed to spending $38 billion over the coming years on infrastructure rented from Amazon’s cloud business. Beyond that, Amazon may be negotiating the use of its Trainium AI chips inside OpenAI’s stack.
Jupiter and Venus — meet SpaceX and xAI
Elon Musk is eyeing a June 2026 IPO for SpaceX — timed to coincide with a rare Jupiter–Venus planetary alignment and his birthday on June 28. Musk is reportedly considering a merger between SpaceX and xAI ahead of the IPO.
And Tesla is in the orbit. Tesla has invested $2B in xAI, positioning the move as part of a broader pivot toward AI and robotics, well beyond pure EV manufacturing.
A crowded release window for AI models in China
ByteDance and Alibaba Group are both preparing to release their next flagship AI models around the Lunar New Year holiday in mid-February.
ByteDance plans to launch new language, image, and video models, while Alibaba is expected to release Qwen 3.5, its next flagship model.
The competition could be even fiercer this year, as DeepSeek is also planning to release its next major model around the same time — its first in more than a year.
Brex and Ramp diverge
Capital One struck a deal to buy payments and credit card startup Brex for $5.15 billion — above Brex’s latest secondary-market implied valuation of $3.77 billion. As of September, Brex was generating more than $700 million in annualized revenue.
Brex’s main rival, Ramp, raised money last summer at a $22.5 billion valuation, followed by another round in November at $32 billion, after taking business away from Brex and other competitors by offering slick software and more credit card rewards.
Funding, deals, and strategic moves
Revolut has reportedly scrapped plans to acquire a U.S. lender and will instead apply for a standalone banking license, betting on a lighter regulatory approach under the Trump administration.
Notion completed a $270 million tender offer at an $11 billion valuation.
Redwood Materials raised $425 million at a $6 billion valuation — more than $1 billion above its prior round — bringing total private capital raised to $2.3 billion. Google joined the cap table for the first time, alongside Nvidia and other strategic investors. The draw in this round appears to be Redwood’s expanding energy-storage business, increasingly positioned as critical infrastructure for powering data centers.
Airtable is entering the AI-agent space with Superagent, its first standalone product in the company’s 13-year history. According to Airtable’s CEO, Superagent could eventually eclipse the core Airtable product.
AI startup World Labs is reportedly raising hundreds of millions of dollars at a $5 billion valuation, up from $1 billion when it emerged from stealth in 2024. The company is focused on “large world models” — AI systems designed to understand and operate in 3D environments. Late last year, it launched its first product, Marble, which generates 3D worlds from text or image prompts.
British startup Synthesia, which helps companies create AI-generated training videos, raised a $200 million Series E at a $4 billion valuation, up from $2.1 billion just a year ago. The company reached $100 million ARR last April, is now at $150 million ARR, and expects to pass $200 million in 2026. Following the round, Synthesia will facilitate an employee secondary share sale.
Clay, an AI sales and marketing startup, is expected to be valued at $5 billion in an upcoming tender offer.
Customer-support AI startup Decagon raised $250 million at a $4.5 billion valuation in a round led by Coatue and Index Ventures.
Paris-based Ledger is preparing for a U.S. IPO that could value the company at more than $4 billion. If completed, it would be one of the first VC-backed European crypto companies to list publicly. The valuation is notably more conservative than recent U.S. crypto IPOs — Circle was valued at $8.1 billion, while Bullish priced at $5.4 billion.
C3.ai, founded by Thomas Siebel, is in talks to merge with privately held Automation Anywhere, a deal that would effectively take Automation Anywhere public.
SambaNova Systems is considering raising up to $500 million after talks to sell the company to Intel stalled. Intel had previously discussed an acquisition that would
Flapping Airplanes, a research lab focused on more data-efficient model training, raised $180 million at a $1.5 billion valuation.
Enjoyed this article?
Subscribe for private market insights delivered to your inbox.