Nvidia Just Warned of 15% Price Hikes. Here's Who Pays the Bill.

Usage-based and hybrid pricing models are changing how B2B companies generate revenue — and creating new headaches for the finance teams behind them.

Tabs co-founder Rebecca Schwartz and PwC Partner Amit Dhir sat down to unpack exactly what that means in practice: how pricing model decisions ripple into revenue recognition, forecasting, and financial ops — and what it takes to scale without piling on manual work.

Watch the on-demand recording to get practical frameworks, real-world examples, and a clear path to operationalizing usage-based revenue — including a forward-looking take on how AI will reshape financial workflows. If your team is navigating pricing complexity heading into the back half of the year, this is worth an hour.

Watch the recording now

.

1. Nvidia warns its biggest customers that AI server prices are going up more than 15%.

Servers built around Nvidia's flagship Grace Blackwell and Vera Rubin chips will cost over 15% more in many configurations, starting with systems shipped early next year, according to a Bloomberg report. The driver is not GPU margins but soaring memory prices, and contract manufacturers building for Microsoft, Google, and Oracle have already notified their customers. Read the CNBC summary or see Tom's Hardware's margin math.

2. Nvidia is in talks to back Perplexity at a valuation above $30 billion.

The equity round under discussion would lift Perplexity's valuation more than 50% from the $20 billion it reached about a year ago, The Information reported. Annualized revenue has climbed from under $250 million at the start of 2026 to more than $750 million, driven partly by Perplexity Computer, its agent that automates desk work for professionals. See the full funding history.

3. Apple announces a new Mac mini built for local AI.

The refreshed desktop comes with a choice of the new M6 chip or the M5 Pro, and the M6 packs a 12-core CPU plus the first dual Neural Engine in Apple silicon. The pitch is clear: run AI models on your desk instead of in the cloud, though the same memory crunch squeezing Nvidia is expected to push the Mac mini's entry price above its old $599 floor. Get the full spec rundown and the backstory on the rushed launch.

The sticker shock. Nvidia has told its largest customers that AI server prices will rise more than 15% in many cases on systems shipping early next year. The company runs gross margins around 75%, among the highest in semiconductors, yet it is passing memory cost inflation downstream rather than absorbing it. Tom's Hardware breaks down why it can.

Blame the memory. Three companies, Samsung, SK Hynix, and Micron, produce most of the world's DRAM, and AI demand has handed them rare pricing power. The squeeze is already spilling beyond data centers, with Apple and Qualcomm both saying chip costs forced their own price rises. See how the suppliers gained the upper hand.

The forecast. Gartner projects the memory shortage will last at least into 2027, while Deloitte expects AI server DRAM prices to quadruple over the full year and says meaningful new supply may not arrive until 2029 or 2030. In other words, this is not a blip. Read the hidden cost analysis.

Who actually pays. Hyperscalers keep buying because demand for compute still outruns supply, so the hikes raise the capital required per data center rather than cooling the appetite. Consumers feel it at the edges, from pricier phones to a Mac mini expected to jump from $599 to as much as $799. See the consumer ripple effects.

The bigger game. Nvidia is converting cost pressure into strategy, using $48.5 billion in quarterly free cash flow to finance its own customers and lock in future demand, what CNBC calls a moat built from capital rather than chips. If compute is the new oil, the winners of this cycle will be decided by balance sheets as much as benchmarks. Read CNBC's take on the capital moat.

Claude reads the room. Anthropic updated Claude Tag, its Slack agent, to read entire channels instead of one message at a time, making it about 30% better at deciding when to jump in unprompted and when to stay quiet. See how Anthropic grades its restraint.

Vera's report card. At Hot Chips 2026, Nvidia opened up its 88-core Vera CPU, showing 1.2 TB/s of memory bandwidth and claiming up to 1.8x faster agentic task completion than x86 chips, a number rivals are already contesting. Get the full technical breakdown.

Grok goes orbital. SpaceXAI will run Grok's AI agents on standalone Vera CPUs and, per an Elon Musk post on X, launch a space-optimized Vera Rubin NVL72 system aboard its first Starmind satellite in late 2027. Watch the plan take shape.

Poolside's payday. Nvidia quietly agreed to pay $6 billion to license AI model development software from startup Poolside while also investing $1 billion at a $12 billion pre-money valuation, one of the largest licensing deals in AI so far. See what Nvidia gets for the money.

Cook's last launch. The new Mac mini is expected to be the final product Apple unveils under Tim Cook before he steps down as CEO, a quiet close to a 15-year run that began with the iPhone 4S. Read why the timing matters.

PROMPT STATION

Turn any price increase email into a negotiation plan

Vendors everywhere are quietly raising prices this quarter, and most people just accept the new number. This prompt works in ChatGPT, Claude, or Gemini and turns a renewal notice into a clear counter-strategy, complete with a ready-to-send reply. Try it the next time one lands in your inbox.

Swap [PASTE THE EMAIL] with the vendor's actual message, a renewal quote, or notes from a pricing call. Try adding your contract end date and annual spend so the counter-asks get specific to your situation.