A Data Center Check-In

Public markets were up modestly over the past week, with the S&P500/Nasdaq100 1%/1%.  YTD the S&P500/Nasdaq100 are now +2/+3%.  Our NPM Private Market Tracker*, which shows the average price performance of the 50 largest names in our internal Tape D® data, is +2% YTD. (NPM; Bloomberg)

The AI data center industry continues to experience an unprecedented build-out, driven by explosive demand.  In private markets data center providers Lambda and Crusoe were two of the best performing names overall in 2025, +199%/+248%, respectively,However, despite periodic “bubble” fears in the market, the Big 5 hyperscalers (AMZN, GOOG, META, MSFT, ORCL) continue to ramp spend.  Based on public guidance these companies will spend ~$600bn on infrastucture in 2026, a roughly 1/3 y/y increase.  (Bloomberg; NPM)

On the GPU economics front, the market has seen volatility on rental pricing for GPU clusters, but chip level economics remain strong.  According to Silicon Data, rental pricing for the high end NVDA B200 is down ~16% over the past year.  Rental pricing, however, on the last gen H100 chip is actually +11% over the same time period. This pricing supports the notion that inference applications are less “generation sensitive” giving older chips perhaps a longer economic life than some had feared.  (Bloomberg; NPM; Silicon Data)

As with any rapid growth market, high returns are attracting new competitors and alternative infrastructure providers. For example, cloud infrastructure provider Vultr is building AMD-powered (vs. NVDA) AI clusters, with pricing that aims to undercut major hyperscalers and cloud providers.  Another example is OpenAI’s recently announced a ~$10 billion agreement with chip start-up Cerebras Systems through 2028, which to our knowledge was the first major AI customer win for the company. (Bloomberg; StrictlyVC)

With rapid growth also comes new challenges.  Power constraints remain among the most significant bottlenecks for new buildouts.  In addition, supply chain pressures around GPUs, high-bandwidth memory and specialized networking components continue to complicate timelines. Sustainability concerns — including electricity consumption, water use for cooling, and local community impacts — are also increasingly attracting regulatory scrutiny, which could influence the cost and pace of future projects. (Bloomberg; NPM)

BIGGEST MOVERS AND TOPICAL NAMES

Based on our proprietary Tape D® data, the best performers of the large cap names in the private market thus far in 2026 have been Cerebras (+32% estimated share price performance), x.AI (+28%), and OpenEvidence (+20%).   (Source: NPM)

RECENT EVENTS

• Tesla announced the end of production of its Model S and X verhicles, with factory capacity being reallocated for humanoid robot and robotaxi production. (1/28; Bloomberg)
• President Trump announced a new plan to hold a special electricity auction for tech companies seeking new power capacity to fuel AI data centers. (1/15; Bloomberg)
• Apple announced that it will be using Google’s Gemini AI on its iPhones. (1/14; StrictlyVC)
• OpenAI announced a deal to buy 750MW of computing power from chip startup Cerebras Systems over 3 years. (StrictlyVC; 1/14)

NOTABLE CAPITAL RAISES

• OpenAI is in talks to raise $100bn, with up to $60bn coming from NVDA, AMZN and MSFT. (1/29; The Information)
• Anthropic is in talks to raise $20bn at a $350bn valuation. (1/27; StrictlyVC)
• Replit is in talks to raise $400mm at a $9bn valuation. (1/15; Bloomberg)
• OpenAI invested $250mm in Sam Altman’s brain computer interface company, Merge Labs, at an $850mm valuation.
• SkildAI raised $1.4bn at a $14bn post money valuation. (StrictlyVC; 1/14)

NOTABLE EXITS

• Thoma Bravo backed software company Anaplan is preparing for a confidential IPO filing. (The Information; 1/28)
• Antrhopic, OpenAI and SpaceX are all considering IPOs in 2026.  (NYT; 1/14)