The Long Line · Series 1 · DAC 2026, Long Beach
You’re holding a piece of the line.
From the 1866 transatlantic telegraph cable to the millimeter between two chiplets, the same engineering instinct has carried a clean signal across a hostile channel for 160 years. These 21 cards trace that line — where it came from, what it’s bringing back, and where it goes next.
Look behind. Look forward. Enable the future.
Five cards in every pack. No pack completes a set.
One card from each suit, a wild for trade currency, and a hit slot — usually a common, sometimes the foil Orchestrator (about 1 in 7). You finish a set the old way: you trade. Find someone on the floor holding what you’re missing.
The Lineage · look behind
Where the instinct comes from. Tap a card to flip it.
01 / 18 · 1866The First Long Lineflip ⟳
The cable ship Great Eastern finished a working telegraph line across the Atlantic. The hard part was never the wire — it was getting a clean signal across 2,000 miles of hostile channel. Lord Kelvin’s mirror galvanometer was built to read a whisper at the far end. Every SerDes equalizer shipping today solves the same problem. The first long line ran under an ocean; the newest run between two chiplets a millimeter apart.
Look behind. Look forward. Enable the future.
02 / 18 · 1947The Device That Started Itflip ⟳
December 1947, Bell Labs, Murray Hill, New Jersey: the first working transistor. Everything in modern computing traces back to a bench on the eastern seaboard. The materials changed; the geography of the expertise — New Jersey, northern England, Scotland — did not. The people who learned to push electrons cleanly are still here.
Look behind. Look forward. Enable the future.
03 / 18 · 1995Right Idea, Wrong Decadeflip ⟳
Intel’s Pentium Pro shipped a CPU and its cache as two dies in one package — a real multi-chip module in a mainstream product. Within two generations Intel went back to one big die; the economics were not ready. The idea was not wrong. It was early. The best ones usually are.
Look behind. Look forward. Enable the future.
04 / 18 · 1956The Dream, Waiting for the Machineflip ⟳
Artificial intelligence was named at a summer workshop in 1956, and the math was mostly there by the 1960s. What was missing for sixty years was compute. AlexNet was 60 million parameters; today’s models pass a trillion. AI is not new — it is a last-century idea that finally has the silicon to run on. What a time to be alive.
Look behind. Look forward. Enable the future.
05 / 18 · 1965The Law That Set the Paceflip ⟳
Gordon Moore wrote four pages in Electronics magazine predicting transistor counts would double on a cadence. It was an observation, not a law — but the industry made it one, spending sixty years and trillions of dollars keeping a promise one man sketched on graph paper. The pace was never physics. It was will.
Look behind. Look forward. Enable the future.
06 / 18 · the lineageThe SerDes Coastflip ⟳
Draw a line from the cable stations of Northern England and Scotland to the labs of New Jersey and the eastern seaboard. For 160 years, the same coasts have produced the people who move signals cleanly across hostile channels — from ocean floors to backplanes to the millimeter between chiplets. Expertise has a geography. It rarely moves.
Look behind. Look forward. Enable the future.
The Reset · old is new again
The technologies the industry called dead ends — now carrying the most advanced silicon on Earth.
07 / 18 · 1980sMulti-Chip Modulesflip ⟳
IBM put multiple dies in one ceramic package to keep its bipolar mainframes running. Niche, expensive, slow — the industry called it a workaround. Forty years later we call it a chiplet, and it is how the most advanced AI silicon on Earth gets built. Old is new again.
Look behind. Look forward. Enable the future.
08 / 18 · 1984 → 2019A Datacenter on a Waferflip ⟳
Gene Amdahl’s Trilogy Systems tried to build a computer on a single uncut wafer and burned through one of the largest startup funds of its era. Wafer-scale integration was declared a dead end. Cerebras now ships exactly that — a processor the size of a dinner plate. The dream did not change. The yield and the tooling finally caught up.
Look behind. Look forward. Enable the future.
09 / 18 · nowPackaging Is the New Lithographyflip ⟳
‘System on a chip’ is becoming a misnomer. Modern systems are not on a chip — they are on a package. The new Moore’s Law lives in how dies are stacked and joined, and the chiplet market is growing about 42% a year through 2030. Look behind: a telegraph cable. Look forward: a composable AI fabric. Same engineering instinct.
Look behind. Look forward. Enable the future.
10 / 18 · UCIePCIe for Chipletsflip ⟳
PCI let any card talk to any motherboard, and an industry bloomed. UCIe is the same bet at package scale: a standard die-to-die interconnect so chiplets from different vendors — different foundries, different nodes — snap together like it’s 1995 again. Open standards don’t just connect parts. They create markets.
Look behind. Look forward. Enable the future.
11 / 18 · HBMMemory Moves Next Doorflip ⟳
For decades, memory lived across the motherboard and compute waited on it — the ‘memory wall.’ HBM knocked the wall down by stacking DRAM and moving it millimeters from the processor on a silicon interposer. Every frontier AI model trains on it. The fix wasn’t faster memory. It was shorter distance — the oldest trick on the line.
Look behind. Look forward. Enable the future.
12 / 18 · mixed nodesEach Function, Its Own Nodeflip ⟳
A monolithic SoC forces everything — logic, analog, I/O — onto one bleeding-edge node, whether it benefits or not. Chiplets end the tyranny: compute on the newest process, analog and I/O on mature nodes that do the job for a fraction of the cost. The right silicon for each function, joined in one package. Not one chip. One system.
Look behind. Look forward. Enable the future.
The Stack · look forward
The six layers of EDA 3.0 — the AI-native lifecycle from intent to yield, and the companies AiT represents at each layer.
13 / 18 · L1Intent & Requirementsflip ⟳
Jama · Glide
Every chip that failed in silicon first failed in a sentence — a requirement misread, an intent lost between the spec and the RTL. The first layer of EDA 3.0 makes intent machine-readable and traceable from boardroom to tapeout. Get the words right and the silicon follows.
Look behind. Look forward. Enable the future.
14 / 18 · L2Architecture & Modelingflip ⟳
Rise · InPsy
Before a single gate exists, the architecture decides the outcome: partition, power, package. This layer explores thousands of what-ifs in software before committing millions to masks. The cheapest bug to fix is the one caught before the design exists.
Look behind. Look forward. Enable the future.
15 / 18 · L3RTL & Firmwareflip ⟳
CraftifAI · Rise
Sixty years after the transistor, humans still hand-write the logic. This layer is where AI stops assisting and starts generating — RTL and firmware produced from higher intent, reviewed by engineers who move from typists to editors. The hands change. The judgment doesn’t.
Look behind. Look forward. Enable the future.
16 / 18 · L4Verificationflip ⟳
Axiomise · Quaxys
Verification consumes most of every chip schedule, and simulation can only show a bug it happens to stumble into. Formal methods prove the absence of entire bug classes — mathematics, not luck. In an era of AI-generated logic, the question isn’t ‘did we test it?’ It’s ‘can we prove it?’
Look behind. Look forward. Enable the future.
17 / 18 · L5Yield & Physicalflip ⟳
YieldWerx · Tuple
A design that yields at 60% is a different business than one that yields at 90% — same RTL, different company. This layer closes the loop between the fab floor and the design desk, feeding manufacturing reality back upstream. Moore’s Law was made in the design office. Margin is made here.
Look behind. Look forward. Enable the future.
18 / 18 · L6AI Orchestrationflip ⟳
the open seat
Every layer below this card has an owner. The layer that connects them — carrying intent from L1 through verification to yield and back — belongs to no one. The incumbents own tools. Nobody owns the line. This seat is open. We know because we’re sitting in it.
Look behind. Look forward. Enable the future.
Two cards you can’t count on pulling.
◆ Foil chase · ~1 in 7 packs ◆
The Orchestrator
You found it. One in seven packs carries this card, which is roughly the odds any given idea in this industry gets its timing right. The orchestration layer — the thread connecting intent to architecture to silicon to yield — is the rarest thing in a $1T industry: unclaimed. Trade this card if you want. We’re keeping the seat.
No. ___ / 250
The Founders’ Card
This card was never in a pack. It was handed to you by Simon Bennett or John Simmons, which means you had a conversation with the people who’ve walked the line. Two hundred fifty exist. Yours has a number and a handshake behind it. That’s the rarest finish we print.
Want one? Find us at booth S-3B.
Your want list
Check off what you’re holding. Whatever’s left unchecked — that’s who you’re looking for on the floor.
The Lineage
The Reset
The Stack
The Chase
The line runs through booth S-3B.
The cards tell the story. The conversation is better in person. Book a session with AiT and our clients at DAC 2026 — group deep dives, 1-on-1s, or Simon’s EDA 3.0 overview talks.
The Long Line, Series 1 · a DAC 2026 giveaway from AI Tech Sales · not for resale · www.ai-techsales.com