Subject: In Huang We Trust
Cantor Fitzgerald
(2/17/26)

"Stay Calm and Trust in Huang as We Remain at the
Cusp of a Rack Scale Supercycle

The story is so unbelievably simple, yet at the same time quite complex
today. Near-term, we expect a strong beat and raise in two weeks, driving
consensus materially higher – we continue to think CY26/27 EPS is tracking
towards $9.00/$12.00+ vs consensus today of $7.75/$9.53. Compute
demand remains insatiable with no line of sight to these constraints easing
(even A100, a 6yr old product, is sold out with rising pricing). More specific
to CY26, we are at the cusp of a rack-scale super cycle as Blackwell
continues ramping and Rubin remains on the come in the back-half of
CY26 (expect messaging to support ramp plan on track). Bigger picture, we
are still in the very beginning of the next industrial revolution driven by
proliferation of Agentic + Physical AI (likely driving the largest buildout in
history). So the setup is extremely positive, particularly when we reflect
on shares only trading 21x/16x expected CY26/27 earnings. On the other
hand, investor concerns remain, headlined by fears of peaking hyperscale
Capex in CY26 (NVDA barely budged on the significant Capex raises from
AMZN/MSFT), rising competition from Google TPU’s, and overall worries
related to circular financing/unsustainable Capex spending. Our view, this
“zero-sum” view is flawed, and the reality is that AI demand is exploding,
and we believe NVDA is now sold out for all of CY26, with backlog now
building for CY27 and CY28. Our sense is that this earnings report will
mark the beginning of an inflection point for shares, followed by the latest
frontier models built on Blackwell released in March highlighting NVDA’s
technological leadership, GTC San Jose in March further highlighting the
company’s leadership in AI, and finally a likely Data Center revenue guide
update for CY27 from Jensen as early as Computex in early June given our
vision for AI lead times to continue to extend given the insatiable demand
for AI Accelerated Compute. With a clear path to EPS tracking to $12+, we
continue to see fair value of AT LEAST $300, which would equate to a 25x
multiple – though given growth prospects through the end of the decade,
we believe an argument could absolutely be made for 30x+, which would
support a bull case closer to $400. NVDA remains a TOP PICK."