Wanted: A few Good Magicians

Current and next-gen interconnect requirements may prompt the need for R&D budgets, because 'legacy-gen' interconnect solutions simply won't work as data rates and densities escalate.

Data.

It is the reason the Electronics Manufacturing Industry exists in the first place.

This seminal element of a massive industry’s origin often gets lost in the signal-to-noise ratio emanating from business development and marketing offices, senior suites and board rooms alike, in a clotted market ecosphere where complicated and expensive efforts to differentiate go-to-market models often emulate the act of throwing dice against a wall.

Customers aren’t just looking for suppliers that know how to solder components onto substrates and wrap metal or plastic around them. They aren’t just looking for suppliers that invest in quality management systems and various certifications meant to engender reliability. They aren’t just looking for proximity advantages or the counterpoint of low labor cost regions, nor even the mastery of complex supply chain management to imbue pricing leverage.

They are seeking all these things, to be sure, of course.

But they are also looking for suppliers that have a comprehensive understanding of the implications behind managing the vast quantities, quality, and diversity of data required to be pumped through systems serving the infinite end-use cases of their own customers.

And what do such end-use customers want?

Here is an instructive anecdote:

When I worked for a premier market intelligence firm some years ago, one of our customers was a world-renowned content creator, aggregator, and distributor in the entertainment space. In my very first meeting with their Strategy/Planning group, I asked them to summarize where and how our analysts could best help them.

One of their team cleared his throat and said, “We have many terabytes of archived analog properties that need to be digitized, digitally watermarked, their digital rights properly managed, and then all of that content has to be distributed anywhere, on demand, all at once, and we need to get paid for their use automatically. We also have to do the same thing with any new properties we’re developing”.

When we dove into that problem set, there were enormous hardware and software implications requiring a global team of analysts to help guide the decisions and investments necessary to solve it.

Scroll forward to the electronics industry of today.

There is no appreciable difference between the problem set of that content creator and, say, the clinical setting that needs to rapidly receive data-intensive native image files from a variety of scanning equipment, convert them to a DICOM standard, integrate them as needed, deploy them remotely and collaboratively, and then analyze the scans to provide effective image diagnostics and patient care, including within the eHR universe.

There is no appreciable difference, either, for transportation systems, like a railway or a ship, needing to completely sensor the rail cars, the tracks, and the stations, or the ships, cargo, and their harbors…to create, harvest, digest, and distribute real-time operational information. Things like the speed, roll, pitch, and yaw of a ship on the open ocean, weather conditions, traffic, or the buoyancy readings that could, for example, indicate contraband acquired or removed en route, and similar information on a rail path.

There is no appreciable difference, as well, for Generals in a battlespace, facing multiple potential real-time adversaries, needing to instantly decide which of them poses the most immediate threats (including simultaneous ones), to sight and acquire them all, decide what countermeasures to use and which of the Services, Agencies and individual soldiers to use, whether to use a legacy or new platform, where they are geo-located, and necessitating Artificial Intelligence (AI) solutions, or perhaps even AI-within-AI solutions.

It’s about the speed of access to ever-multiplexing, complex, actionable data and decision-making and it’s about the quality and security of the signal integrity needed to evolve and curate the data across time and space as conditions and demands exponentiate.

Waxing political for a moment, "It's the economy, stupid" was a celebrated motto of Bill Clinton's first presidential campaign and attributed to his campaign strategist, James Carville, in 1992. Carville was emphasizing the importance of the struggling economy in their campaign messaging and he boiled a pile of political intellect and doublespeak down to a folksy, direct phrase people could understand.

Let’s consider a different kind of campaign.

This one has a few hundred company presidents, all strategizing how to best differentiate their companies in a clotted field. In the end, scads of money get thrown at various methods for how to ‘chase the same nickel’ through marketing initiatives, consolidation activities like mergers and acquisitions, and threading the prognostication needle for Wall Street.

I am referring, of course, to the $400-500B global Electronics Manufacturing Services industry.

I am also referring to the timeliness for a similarly relatable, poignant message to the one Carville issued, amid the ‘-tronics this' and '-tronics that’, or ‘Hi-Rel this' and ‘Hi-Rel that’ hyperbole, wrapped in branding statements like: ‘Made Possible. Made Better’, ‘Partnering with you to…’, etc.

Whatever.

“It’s the Data, stupid” could be a candidate, but it’s deeper than that.

The exponentiating demands for data have, for the last forty years or so, tracked along the Moore’s Law rubric: the number of transistors in an integrated circuit (IC) doubles about every two years.

In a Military Microelectronics conference that I spoke at ten years ago, there was consensus from all participants – including NSA, IBM, AFRL, DARPA – that ‘More Than Moore’ was at our doorstep then.

In one case, a participant delivered a presentation positing the concept of carbon nano I/O switches to replace standard transistor models and increasing on-chip densities by orders of magnitude.

That same year, DARPA issued a Broad Agency Announcement (BAA) asking for proposals to help develop the Electronic-Photonic Heterogeneous Integration (E-PHI), a solution “[…] providing considerable performance improvement and size reduction versus current, state-of-the-art technologies”. E-PHI essentially involved a processor layer, a memory layer, and a photonics layer inside the same chip.

Today, the next generation DARPA BAA’s include the CHIPS (Common Heterogeneous Integration and IP Reuse Strategies) program: Integrating digital, analog, and mixed-signal blocks into system-on-chip (SoC) solutions, and the PIPES (Photonics in the Package for Extreme Scalability) program: Replacing traditional I/O with optical signaling interfaces to achieve major improvements in link reach and efficiency.

Why the need for these ground-breaking initiatives?

Because accommodating the speeds and volumes of data creation, acquisition, transmission, distribution, and digestion simply won’t work with conventional materials, assembly footprints, or manufacturing techniques and chip-to-chip interconnections must now operate at ‘on-chip’ speeds.

That has huge implications for the several hundred EMS executives.

We have now arrived at ‘Moore’s Inflection’, described in a brilliant article published by Defense Media Network in 2018, by Dr. William Chappell, Director of DARPA’s Microsystems Technology Office (MTO). In that article, he shared this chart ( https://www.defensemedianetwork.com/stories/the-intertwined-history-of-darpa-and-moores-law/ ):

That inflection point implies some curious possibilities worth considering:

*    The very differentiation that EMS companies are clamoring to achieve with marketing budgets or M&A strategies, might otherwise be achieved by taking the same money and proactively leading the marketplace with novel approaches to solving next generation interconnect problems before being asked by customers.

(Note Chappell's chart: Specialization and Uniqueness are the future. Generalization and Abstraction are hallmarks of the last 40 years and we can - and should - begin saying 'goodbye' to those).

*  There is generally a financially driven, risk-averse, collective mindset permeating the electronics industry ever since the halcyon days when R&D budgets were a staple in printed circuit board manufacturing shops. Those budgets, that arguably underpinned an essential reason the industry grew substantially in the '80's and '90's, were relegated to the dustbin in the face of overseas competition. The ‘dot.bomb’ anomaly in 2000 and the financial melt-down of 2008 exacerbated any remaining fears about risk, and COVID-19 has only amplified their intensity.

*  Perhaps it is time for companies to take some new, calculated risks and retrieve the notion of R&D again, becoming true champions for next generation data transmission and ‘sandboxes’ for customers to play in, where collaborative ideas and solutions can be developed. This means making standalone budgets aimed at critical signal management techniques and other development, not ‘pay-to-play’ models like Creative Workshops where most customers lack discretionary budgets to pay for idea-sharing.

* There are also some simple facts arriving at any EMS shop near you soon:

- The need for circuit card footprints that surface mount and other equipment were geared to accommodate in bygone years are becoming just that: Bygone. Think instead about the nature of a 14mm X 14mm, low Dk non-ceramic substrate, with 5μm lines and spaces, additive technologies, integrating analog, digital, and mixed-signal components, and photonic interconnect that multiplexes an electronic signal into light waves and de-multiplexes the waves back into an electronic signal, with zero signal loss, up to meters in transmission length. Imagine that as a 'norm'.

-  The need for standard organic substrates is also, selectively ‘bygone’. The simple fact is that conventional materials cannot withstand very high frequencies (60-100+Ghz) and copper won’t properly conduct the signal without extremely short transmission lines. This ushers-in the ideas of stacked die and integrated photonics, as well as orders of magnitude smaller, denser real estate.

-  The price points for such packages are also orders of magnitude smaller per unit than what an industry, steeped in circuit card assembly and box-build assumptions, can otherwise fathom. This implies an industry, in the throes of Moore’s Inflection, that now has to grapple with how to evolve on the fly, how to acquire and apply new knowledge, how to articulate better value propositions (without the hype), how to educate investors, and how to maneuver and sustain relevance as data demands surmount past models and assumptions.  

-  Trust is everything. There is a reason why certain OEM’s and Defense Primes have maintained their captive manufacturing and assembly capabilities. In the case of Primes, they are expanding their internal capabilities as assembly packages shrink. The nature of their end-use cases involve increasingly sensitive data and are, among other things, subject to signal intrusion. So, protection of signal integrity in the end-use applications is crucial to maintaining advantages against competition and foes alike.

- The field of available, competent, trusted suppliers has simply not kept pace with demand. This is almost certainly due in no small way, to a deficit in the electronics supply chain for proactively understanding why certain applications are developed in the first place, let alone prescribing new solutions that make them a reality. (See also R&D above).

Moore’s Inflection is a decision point for the entire EMS industry. Waiting for customers to tell the supply chain what to do is a cousin of the adage: ‘Rest assured that your recent accomplishments at work have gone largely unnoticed’.

Arthur C. Clarke once said, “Any sufficiently advanced technology is equivalent to magic”.

What the electronics manufacturing industry needs now – to differentiate, to be trusted, to survive an unalterable evolution, to be in fact ‘sufficiently advanced’ – are a few magicians taking risks and setting the pace for decades ahead.

Those magicians are typically found in research and development labs, pioneering solutions that the world inevitably needs. Customers have these, why not the greater supply chain? That could become a great vehicle for joining forces and partnering, thereby strengthening trust, value, and differentiation.

The EMS industry has the opportunity to lead in the face of huge, unavoidable changes to the interconnect landscape ahead.

The question then becomes: Will it?

Steve DeWaters