RS

Technology investor

[RS @_Reeesh]

Spot on. all the value is captured early, leaving retail to hold the bag.

Recent IPOs perfectly expose this private vs public price disparity. Mega VC funds inflate late stage valuations to sizes that cant survive public market scrutiny

No wonder tech stays private longer the bar to exit has become astronomically high

[RS @_Reeesh]

@MorningBrew ahh yes, I remember the absolute economic wasteland of the 1990s and mid-2000s when everyone was famously broke

[RS @_Reeesh]

Cerebras (solana:4yEjcMiy6GAgrpWpUvhUXfaP1vQmJXfqJjEyxBSZpump) indicated to open ~$360. That is ~$110B market cap on $510M of 2025 revenue.

216x trailing P/S. ~48x on 2027E (est. $2.3B).

For reference: $NVDA at its peak AI-froth multiple in mid-2024 traded around 40x trailing. At that point NVDA was a $100B< revenue business growing 200%+ YoY with the CUDA moat fully baked in.

CBRS: $510M revenue, 76% growth, two-customer base. The market is paying more for projected revenue than it paid Nvidia for delivered.

Customer concentration is also not being priced in. Cerebras pulled its 2024 IPO because 80% of revenue came from one customer. In 2025, 86% came from two UAE entities (G42 + MBZUAI). Going forward, the story flips to OpenAI: $24.6B backlog, majority tied to one buyer

Of that backlog, only $3.7B is contractually scheduled across 2026 + 2027. The remaining ~$21B stretches through 2028 - 2030+ and depends on physical infrastructure being built

The OpenAI relationship is also circular. OpenAI is an equity holder AND lent Cerebras $1B in working capital against the chip commitment

At this price, you are underwriting: flawless execution by Cerebras and its customers, no macro deterioration, 30-40% beats on growth and guidance, and material unforeseen upside on top.

[RS @_Reeesh]

Great data point. I would bet we’re early in adoption rn and after 12 months, looking back at p&ls, uncapped compute spend without clear demonstrable roi becomes a real point of convo in boardrooms. Lots of employees are being told to token max and experiment. This is going to prove very costly over time and CIOs are going to have to optimize this.

[RS @_Reeesh]

@NovusOrderM Objectively The single digit fwd p/e signal as a no brainer has been a mistake many have made in the past.
Today's counterargument is more that there has been a structural realignment in the memory buying circle, which I can buy

[RS @_Reeesh]

@NovusOrderM This isn't the first time Micron ($MU) has traded at a single digit fwd P/E ratio.

The Micron trade has always been "buy when Fwd P/E is high and sell when its low" due to buying cycles.

Investors have been wary that we are in cyclical peak, as we seen many times before.