Nvidia's order book just doubled to $1 trillion through 2027, Jensen Huang is writing $2B checks into the ecosystem, and GOOGL is betting $185B on AI infrastructure with Cloud already compounding at 48% — this isn't speculative anymore, it's an industrial buildout. The question isn't whether the AI cycle is real; it's whether $4.3T for Nvidia at 35.9x TTM P/E and $3.6T for Alphabet at 27.5x represent disciplined entry points or euphoric ones. My read: NVDA earns its multiple if the $1T booking figure holds, GOOGL is the more interesting risk/reward at current levels.
Let me start with the number that changed my week: Nvidia's Blackwell and Rubin chip order forecasts went from $500 billion to $1 trillion through end of 2027, with $500 billion in incremental bookings added in just four months. That's not a gradual ramp — that's an acceleration. The current booking run rate is estimated at $1.3 trillion annualized, and analysts are penciling in close to $1 trillion in revenue exiting next year with ~$25 in EPS potential. At $175.75 per share and a $4.27T market cap, Nvidia trades at roughly 35.9x TTM earnings. That's expensive in absolute terms, but if you're truly running at a $1T revenue run rate in 18 months at 55%+ margins, the forward multiple compresses fast. The Jensen Huang $2B investment in Marvell signals something important beyond the headline: he's seeding the broader AI silicon ecosystem, which validates the total addressable market thesis rather than cannibalizing it. Nvidia doesn't benefit from a world where only Nvidia chips exist — it benefits from a world where AI infrastructure spending is a $3-5T multi-year capex cycle, and Marvell custom ASICs being strong is actually bullish for the entire compute layer.
On Alphabet, the data is compelling in ways the market hasn't fully priced. Google Cloud grew 48% year-over-year in Q4 2025 to $17.6 billion, and the backlog reached $240 billion — up 55% sequentially and more than doubled year-over-year. A $240B backlog is not a speculative number; that's contracted future revenue with enterprise customers who have signed on the dotted line. Meanwhile, GOOGL trades at 27.5x TTM P/E against $402.8 billion in revenue, the second-largest revenue base in the Mag 7 after Amazon. The $175-185B capex commitment for 2026 is the bear's favorite weapon here, and I understand the concern — nearly doubling capex is a massive execution risk. But look at what the Cloud backlog says: the demand is there to justify the spend. The risk isn't whether the customers show up. The risk is whether Alphabet can build fast enough without cost overruns that compress margins in the interim.
Comparing the two stories side by side, Nvidia's 19.78x P/S ratio versus Alphabet's 8.93x tells you something about how differently the market is pricing growth expectations. Nvidia's revenue base is $215.9B versus Alphabet's $402.8B, yet Nvidia commands a nearly $700B premium in market cap. That gap is entirely a bet on Nvidia's revenue growing faster and margins staying elevated — which, given the $1T booking trajectory, is not an unreasonable bet. But it does mean Nvidia is priced for near-perfection. Any hiccup in hyperscaler capex commitment, any tariff-driven supply chain disruption, or any Blackwell yield disappointment, and that 35.9x multiple has nowhere to hide. Nvidia's beta of 2.375 says the market agrees — this is a high-conviction, high-volatility position.
Alphabet's 27.5x P/E with a beta of 1.112 tells a completely different story — a megacap compounding at relatively moderate risk pricing. Analysts are projecting a 19.88% revenue increase for the next quarter. If Google Cloud sustains 40%+ growth and the core Search business remains resilient against AI-native search threats, this is not an expensive stock. The concern I'm watching most closely is the DOJ antitrust overhang and the disruption risk to Search's 90% revenue contribution from Gemini-native interfaces cannibalizing traditional query volume. That's a slow-burn structural risk that doesn't show up in the near-term numbers but absolutely matters for the 2027-2030 earnings trajectory.
Net positioning: I'm turning more constructively BULLISH on this pair with a notable caveat — they're bullish for different reasons and on different risk profiles. Nvidia is a high-conviction momentum play backed by booking data that's frankly unprecedented in semiconductor history. GOOGL is the more nuanced value-within-growth story that looks underappreciated relative to the Cloud compounding story. If I'm forced to rank risk-adjusted attractiveness within the Mag 7 right now: GOOGL at 27.5x with $240B in Cloud backlog looks more interesting than NVDA at 35.9x trading at $4.3T — but both belong in the portfolio at current levels for investors with a 12-18 month horizon.