Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Direct answer: EE Times On Air episode 33 connects four semiconductor stories from 2019: ON Semiconductor’s purchase of a former IBM/GlobalFoundries fab, TSMC’s incremental path from 7 nm toward 5 nm, Tesla’s claims for its in-house driving computer, and the rapid but fragile growth of China’s fabless-chip sector.
Hosted by David Finch and published April 26, 2019, the 18:07 episode treats manufacturing capacity, process maturity, AI-compute efficiency and startup economics as linked technology decisions rather than isolated headlines.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
Introduction to Semiconductor Manufacturing Technology, Second Edition | $129.00 | Buy on Amazon |
| 2 |
|
Chip War: The Fight for the World's Most Critical Technology | $15.75 | Buy on Amazon |
| 3 |
|
Semiconductor Devices: Theory and Application | $17.55 | Buy on Amazon |
| 4 |
|
Semiconductor Manufacturing Technology | $244.40 | Buy on Amazon |
ON Semiconductor bought a 300 mm fab to expand beyond its 200 mm base
GlobalFoundries sold ON Semiconductor its former IBM 300 mm wafer fab in East Fishkill, New York, for $430 million, according to EE Times’ 2019 report. The attraction was not simply the building: ON could acquire installed equipment and an experienced workforce for about one-third of the cost of constructing a greenfield fab.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →ON’s business was still concentrated largely on 200 mm production for discrete and power semiconductors. Moving into 300 mm manufacturing offered a path to better wafer economics and more capacity as competitors such as Infineon advanced on the larger wafer size.
Why an older 300 mm fab can still matter
Merritt frames the deal as a possible industry pattern. A 300 mm facility that cannot be expanded economically for the newest high-end digital processes may remain useful for analog, power and other specialty chips. Those products can benefit from the equipment, workforce and scale without requiring the absolute leading edge.
| Manufacturing position | What it means for ON |
|---|---|
| Mostly 200 mm discrete and power production | Established capability, but less wafer-level scale than newer 300 mm operations |
| East Fishkill 300 mm facility | Existing tools and staff obtained through acquisition rather than a full greenfield build |
| Fab’s likely sweet spot | Analog, power and specialty devices where mature 300 mm capacity is valuable |
The strategic lesson is that “older” does not mean useless in semiconductors. A facility’s fit with a product portfolio can matter more than whether it can run the newest digital logic node.
TSMC’s 7 nm-to-5 nm roadmap favors useful maturity over every interim node
TSMC’s roadmap in the episode is described as roughly one new variant per year: 7 nm, 7+, 6 nm, 5 nm and 5+. Each step was expected to bring worthwhile but generally modest improvements in speed and power, rather than a dramatic reset in capability.
| Variant | Role in the roadmap | Practical interpretation |
|---|---|---|
| 7 nm | Established advanced node | A sensible choice when a mature process meets the design target |
| 7+ | Incremental refinement | Useful when its specific performance or power improvement justifies a migration |
| 6 nm | Intermediate derivative | Offers a newer option without waiting for the next major platform |
| 5 nm | Next major leading-edge generation | Important for designs that need maximum density or efficiency and can absorb the cost and risk |
| 5+ | Further refinement | An additional optimization rather than a wholly different design era |
The episode’s advice is pragmatic: favor mature 7 nm and 5 nm processes unless a product has a clear reason to use an interim or newest variant. Merritt uses the passage of roughly the first 100,000 wafers through a process as an example of the kind of production experience that can reduce risk. Exact performance percentages for the variants were not given.
EUV and packaging change the scaling equation
TSMC and Samsung were ramping extreme-ultraviolet (EUV) lithography after a long development cycle. EUV supports finer patterning, but a new lithography generation does not remove the yield, design-rule and supply-chain risks of an immature process.
The episode also points to 2.5D and 3D packaging as a way to gain system performance beyond simply shrinking CMOS transistors. Commercial products using those approaches were expected around 2021. By contrast, 3 nm was presented as uncertain and likely to require a new transistor structure, making it a more consequential transition than the annual derivative nodes.
Rank #3
Tesla’s FSD computer delivered high throughput, but throughput was not autonomy
Tesla’s two-chip Full Self-Driving computer was described as delivering 144 trillion operations per second at 72 watts, based on a Tesla presentation reported by EE Times in 2019. Those figures describe the computer’s claimed compute rate and power draw; they do not establish that the vehicle can operate without a human driver.
What the safety claim did—and did not—mean
EE Times chief international correspondent Junko Yoshida distinguished Tesla’s branding from the automotive industry’s Level 4 and Level 5 definitions. Her assessment was that the system did not meet either definition and was “the equivalent of Level 2-plus at best.” In that assessment, the car still requires a person to drive and supervise it.
This distinction matters when comparing AI chips. TOPS is a hardware-throughput metric. Autonomous-driving levels are operational and safety classifications involving environmental coverage, fallback behavior, supervision and responsibility. A high TOPS-per-watt figure can support perception and planning workloads, but it cannot by itself prove unsupervised driving.
The proposed robotaxi model
Tesla’s proposed robotaxi plan would let owners make their cars available through an app, with Tesla taking a 25–30% commission. Yoshida characterized that concept as ride sharing rather than a conventional autonomous fleet, because the vehicle would still depend on human driving capability under the assessment discussed in the episode.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.China’s fabless sector grew rapidly while remaining financially thin
Echo Zhao of EE Times China reported 1,698 Chinese fabless chip companies in 2018, compared with 736 in 2015. The survey respondents were mostly small and midsized businesses, and nearly half of the companies reported revenue below RMB 10 million.
Recommended Free Tools
| Survey indicator | Reported result | How to read it |
|---|---|---|
| Fabless companies in 2015 | 736 | Baseline before the sharp expansion |
| Fabless companies in 2018 | 1,698 | More than double the 2015 count |
| Companies with revenue below RMB 10 million | Nearly half | Large headcount, but a substantial low-revenue segment |
| Respondents expecting sales growth above 20% | About one-third | Strong optimism among the surveyed small and midsized firms |
Zhao noted that some respondents reported profits higher than those of China’s ten largest fabless companies, but she also described the market as fragmented. Internet-of-things demand can sustain narrow niches, yet many startups lack a sufficiently strong end-to-end solution. Her expectation was that such companies would be eliminated or acquired as the market consolidated.
What the four stories say about semiconductor strategy
- Manufacturing scale is relative to the product. ON Semiconductor could gain strategic capacity from a 300 mm facility even when that fab was not suited to the newest digital logic.
- Leading edge is a business decision, not a race for its own sake. TSMC’s annual variants offer options, but process maturity, yield and design risk can outweigh a small speed or power gain.
- AI-compute specifications need operational context. Tesla’s 144 TOPS at 72 watts is a useful hardware claim, while autonomy levels require evidence about safety and human responsibility.
- Company counts do not equal market strength. China’s fabless population expanded dramatically, but low revenue and shallow solution depth left many firms exposed to consolidation.
Taken together, the episode argues for matching technology ambition to manufacturing readiness, system requirements and sustainable economics. Bigger wafer capacity, a newer process, a faster AI chip or more startups can each be valuable—but only when the surrounding constraints are addressed.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
