Fitih "Fitz" Cinnor

Founder & CEO @semiAIfoundry, Ph.D./MBA | x-Intel tech lead, x-3M portfolio director | Semis, MatSci, AI for science. Back after a decade. Building in public.

[Fitih "Fitz" Cinnor @FitzCinnor]

Excited to introduce Semiconductor Pulse by semiAIfoundry
The semiconductor industry doesn’t move in lockstep, even when headlines suggest otherwise. Pulse connects monthly revenue signals from Taiwan’s ecosystem with global quarterly results to uncover shifts in growth, momentum, and profitability.
https://t.co/9lAgLnjjlg
Explore the signals, understand the divergence, and act with foresight to win the future.
#Semiconductors #MarketIntelligence #IndustryInsights #SemiconductorPulse #semiAIfoundry

[Fitih "Fitz" Cinnor @FitzCinnor]

Tbh, you don't really need the group to advise you on the "review and communication of emerging results," nor do you need their help to "assess their significance" or "coordinate their dissemination." Call it what it actually is: a committee for adaptation to AI vertigo. https://t.co/IsW9AdMGjx

[Fitih "Fitz" Cinnor @FitzCinnor]

When a mind of Terence Tao's caliber struggles to see past localized friction, it underscores where we actually are. The singularity isn't a blinding flash of light; it's a fuzzy line of systemic cognitive vertigo and disorientation.

For centuries, progress relied on a natural "difficulty landscape," a gradient of open problems that forged our conceptual machinery. Relentless algorithmic scaling is steamrolling that topology. AI decouples the truth of an outcome from the human understanding required to reach it, delivering the answer while erasing the journey.

The pushback isn't a failure of intellect as some suggest. It's rational grief over lost friction. The sharpest minds are watching a multi-trillion-dollar build-out strip-mine the supply of open problems, experiencing the shock of a horizon outrunning our cognitive refresh rate.

We have entered a world of abundant discovery, and accelerated adaptation is the only path forward.

https://t.co/kyzeHXEbSs

[Fitih "Fitz" Cinnor @FitzCinnor]

@pequityresearch A 3.6% dip after an 87% run is a flag, not a cycle break. August Taiwan memory participation is still broad. Export unit price mixes product and kit. One month as a watch item to be confirmed on wafers, mix, and the next monthly print.
https://t.co/ECDRcuqcew

[Fitih "Fitz" Cinnor @FitzCinnor]

Excited to introduce Semiconductor Pulse by semiAIfoundry
The semiconductor industry doesn’t move in lockstep, even when headlines suggest otherwise. Pulse connects monthly revenue signals from Taiwan’s ecosystem with global quarterly results to uncover shifts in growth, momentum, and profitability.
https://t.co/9lAgLnjjlg
Explore the signals, understand the divergence, and act with foresight to win the future.
#Semiconductors #MarketIntelligence #IndustryInsights #SemiconductorPulse #semiAIfoundry

[Fitih "Fitz" Cinnor @FitzCinnor]

August Taiwan tape still says participation is wide: memory median +256% YoY across the set, assembly and test +30%. Revenue and bits remain different objects. Hyperscaler eSSD can lift ASPs while consumer stays weak. That split is why we read breadth, not one blended price.
https://t.co/JYcj0dmtym

[Fitih "Fitz" Cinnor @FitzCinnor]

Excited to introduce Semiconductor Pulse by semiAIfoundry
The semiconductor industry doesn’t move in lockstep, even when headlines suggest otherwise. Pulse connects monthly revenue signals from Taiwan’s ecosystem with global quarterly results to uncover shifts in growth, momentum, and profitability.
https://t.co/9lAgLnjjlg
Explore the signals, understand the divergence, and act with foresight to win the future.
#Semiconductors #MarketIntelligence #IndustryInsights #SemiconductorPulse #semiAIfoundry

[Fitih "Fitz" Cinnor @FitzCinnor]

258 of 336 names with history grew in each of the last three months. Pulse is the between-earnings canary. Digest is the week’s judgment.
Dashboard, fixed release, and PDF:
https://t.co/PrxhFGS75T
Not investment advice. Public tape. Check the primary filings before you act.

[Fitih "Fitz" Cinnor @FitzCinnor]

Excited to introduce Semiconductor Pulse by semiAIfoundry
The semiconductor industry doesn’t move in lockstep, even when headlines suggest otherwise. Pulse connects monthly revenue signals from Taiwan’s ecosystem with global quarterly results to uncover shifts in growth, momentum, and profitability.
https://t.co/9lAgLnjjlg
Explore the signals, understand the divergence, and act with foresight to win the future.
#Semiconductors #MarketIntelligence #IndustryInsights #SemiconductorPulse #semiAIfoundry

[Fitih "Fitz" Cinnor @FitzCinnor]

August Taiwan tape: 379 of 434 ecosystem reporters grew year over year. Median +35.6%. Equal company weight, not market-cap weight. Participation is wide. Three layers are growing together: core semiconductors 85%, production adjacencies 93%, hardware demand 84%. Memory median is +256%. Foundry +31%. Equipment +37%. Price, mix, and wafers are still different objects.

[Fitih "Fitz" Cinnor @FitzCinnor]

@paulg Are the labs genuinely surprised by the exponential curve they ushered in? Or, did competitive dynamics force raw capability scaling at the expense of defensive engineering? Or, is controlling an intelligence explosion inherently an impossible task?

[Fitih "Fitz" Cinnor @FitzCinnor]

When a mind of Terence Tao's caliber struggles to see past localized friction, it underscores where we actually are. The singularity isn't a blinding flash of light; it's a fuzzy line of systemic cognitive vertigo and disorientation.

For centuries, progress relied on a natural "difficulty landscape," a gradient of open problems that forged our conceptual machinery. Relentless algorithmic scaling is steamrolling that topology. AI decouples the truth of an outcome from the human understanding required to reach it, delivering the answer while erasing the journey.

The pushback isn't a failure of intellect as some suggest. It's rational grief over lost friction. The sharpest minds are watching a multi-trillion-dollar build-out strip-mine the supply of open problems, experiencing the shock of a horizon outrunning our cognitive refresh rate.

We have entered a world of abundant discovery, and accelerated adaptation is the only path forward.

https://t.co/kyzeHXEbSs

[Haider. @haider1]

Mathematician Terence Tao:

"we have to slow down AI. the pace is insane, and there's no reason to be this fast — no reason at all"

It's amazing how willing we are to change everything without any idea what happens afterward

These are extremely nonlinear dynamics

[Fitih "Fitz" Cinnor @FitzCinnor]

Many will focus on the headline CapEx, but the key test is what @altcap showed at @theallinpod summit. If AI revenues track the physical build-out, the trend is intact. Look at the compute footprint exploding from 18 GW to 115 GW by 2028 against a multi-trillion TAM.
Source: https://t.co/4HWLB4G3eq

[Fitih "Fitz" Cinnor @FitzCinnor]

The "stochastic parrot" critique was a failure to appreciate the exponential curve. Critics looked backwards and mistook the flat tail for a hard ceiling, underestimating the emergence that happens when a multi-trillion-dollar build-out meets scaling laws. Brute force or not, the capabilities unlocked in just a few short years are undeniable.

[Fitih "Fitz" Cinnor @FitzCinnor]

Agentic workloads put the CPU back on the die. Matrix throughput is solved, but agentic execution isn't a SIMT problem. It's a branch prediction, memory latency, and state-synchronization problem. When the bottleneck shifts from GEMM to dynamic control flow, raw GPU FLOPs turn into expensive idle cycles.

[Fitih "Fitz" Cinnor @FitzCinnor]

@haider1 The ultimate irony is resistance isn't just futile, it's the fuel for the exponential acceleration. Every attempt to resist simply becomes high-quality input for recursive self-improvement.

[Fitih "Fitz" Cinnor @FitzCinnor]

Everyone tracks the wafer-start projections, but doubling HBM4 output isn't a bit problem initially. It’s a massive thinning and warpage problem. 12-Hi stacks live or die on the glass carrier, the physical grind, and the vias. Yield compounds across fragile mechanical interfaces, which is exactly why physical realization geometry has to be a primary architectural constraint, not just an afterthought.

[Fitih "Fitz" Cinnor @FitzCinnor]

Compressing chip design from two years to three months will be possible, and AI & next-gen EDA are strong at exploring the space. They still hit a wall at physical commitment. The architecture has to stay visible before package and memory freeze. We built that layer so a faster loop does not paint you into a corner on the later pass. https://t.co/Kx3GW0Stgb

[Fitih "Fitz" Cinnor @FitzCinnor]

CRG Systems v1.5.0 is live.
I’m back on this platform after nearly a decade, and ready to engage.
CRG introduces a Shannon-like abstraction layer between computation and physical realization: a theoretical breakthrough addressing a 55-year-old problem in chip-design commitment, now realized as a deployable system grounded in formal theory, reproducible validation, and independently replayable design-space reduction.
Advanced AI is opening new ways to take on enduring scientific and technical challenges. I’m excited to bring this work into the conversation.
Why it matters:
https://t.co/cJHUiT6vfd
Explore v1.5.0:
https://t.co/HUkws3EhFW

[Fitih "Fitz" Cinnor @FitzCinnor]

Semiconductor Executive Digest (19-Sept, 2026)

AI gets the headlines. The semiconductor race is reshaping what’s possible, what it costs, and who captures the value. https://t.co/3hYp5y3yWH

[Fitih "Fitz" Cinnor @FitzCinnor]

For all the alarmism about AI risks, it’s glaring that only 6% of Anthropic’s R&D compute goes to safety. Even acknowledging that safety research requires less compute than training frontier models, a single-digit allocation reveals the true priority (resources overwhelmingly skews toward scaling capabilities, not securing them).

[Fitih "Fitz" Cinnor @FitzCinnor]

Semiconductor Executive Digest — 17 Sep, 2026

AI is broadening the supply chain. It is not yet a volume recovery. Map process and function, not “everything AI.”

Full scan: https://t.co/cmDkZNQPT9

[Fitih "Fitz" Cinnor @FitzCinnor]

Intel CEO Flags AI Supply Squeeze: Memory Prices Up 5–7x, Its CPUs Meet Just 50% of Demand

[Fitih "Fitz" Cinnor @FitzCinnor]

@thsottiaux Neither, ideally a collaborator.

[Fitih "Fitz" Cinnor @FitzCinnor]

Frontier Anxiety Syndrome? Appreciate you voicing this so accurately. Naming the treadmill makes it much easier to step off of it.

An endless, recursive escalation ladder. The tactical, strategic, personal, and existential all come with a trap and a shift required to adapt in this age of exponential acceleration:

1. Tactical (Obsolescence Anxiety)
The Trap: "Is my 10-year skill set expiring in 6 months?"
The Shift: Skills aren't expiring; execution friction is. What feels like obsolescence is the uncomfortable transition from manual doing to hyper-leveraged judgment and taste.

2. Strategic (The Moat Paradox)
The Trap: "Every startup is one API update away from death."
The Shift: Static code was never a permanent moat. Defensibility in rapid-shift eras belongs to velocity, distribution, deep domain trust, and real-world integration.

3. Personal (Milestone Inflation)
The Trap: "Normal life goals now require lottery-level outcomes."
The Shift: The frontier is no longer single-hub bound. High leverage allows you to decouple value creation, lifestyle, and security from local hyper-inflation.

4. Existential (Post-Success Void)
The Trap: "I made the money, but lost the blueprint for meaning."
The Shift: Clearing material hurdles doesn't erase purpose—it removes the excuse of survival so you can finally build what actually matters.