The Transmeta Story

Madmedic
9 min readOct 15, 2018

The Transmeta Story

Let’s go back in time to the year 2000, the Pentium 3 was Intel’s flagship CPU, AMD was racing to 1 GHz with the Athlon, and the general attitude was the sky is the limit. While it came up a few years after 2000, people thought the Pentium 5 would hit 5 GHz, but in short, they were hot, power-hungry machines racing to get hotter and use more power. In that environment, Transmeta came out and announced the Crusoe, a chip that was meant for small devices — to conserve battery life, to reduce heat — and maybe most remarkable of all full Intel compatibility. Let’s look back on the little company that shook Intel to its core, changed the conversation about processors, and maybe could have been ARM before ARM, and how they ultimately came up short.

Transmeta Timeline

The Founding

In 1995, Bob Cmelik, Dave Ditzel, Colin Hunter, Ed Kelly, Doug Laird, Malcolm Wing, and Greg Zyner found Transmeta. At it’s start, it was the definition of a stealth startup, leading to tons of early Internet speculation, particularly after they hired famed Linux creator Linus Torvald in 1997. Despite having the staff sign over 200 NDAs, inside information did start leaking from the vault. In a December 1998 column of PC Mag, JC Dovrak noted:

“There’s been a buzz in the valley and elsewhere about Transmeta, a start-up that’s going to join in the chip fray. It seems that with the help of some old Cyrix honchos, the company is going to make yet another x86 clone. But this one is going to be ultra-low power and designed for modern notebooks.”

And then in January of 1999, an EETimes article wrote:

“Initial word had Transmeta at work on a Power PC clone. Then, the buzz was that the company was designing a Java chip aimed at the nascent market for low-cost network computers. Now, it seems that Transmeta’s effort is focused more on an x86 alternative that boasts either low-power, multimedia or network-computer capabilities. Or perhaps all three.”

Despite a trickle of leaks, Transmeta itself stayed silent until a comment in the HTML of their website finally announced:

Yes, there is a secret message, and this is it: Transmeta’s policy has been to remain silent about its plans until it had something to demonstrate to the world. On January 19, 2000, Transmeta is going to announce and demonstrate what Crusoe processors can do. Simultaneously, all of the details will go up on this Web site for everyone on the Internet to see. Crusoe will be cool hardware and software for mobile applications. Crusoe will be unconventional, which is why we wanted to let you know in advance to come look at the entire Web site in January, so that you can get the full story and have access to all of the real details as soon as they are available.

The Transmeta Crusoe

Slide from Crusoe launch deck

As promised, on January 19, 2000, Transmeta held the Crusoe announce event.

For an in-depth technical look at the Crusoe, Ars Technica did an excellent write up when it was released linked to here. At a high level, this processor used a virtual machine to provide an x86 emulation layer using a system they called “Code Morphing Software.” In theory, this system would also allow them to emulate other instruction sets such as PowerPC. The goal of the Crusoe was to provide full x86 compatibility with acceptable performance while using much less power than Intel and AMD chips.

Riding high from their announcement in January, Transmeta had their IPO on November 7, 2000, and it opened at $21 a share before peaking at $50.26 and ending the day at $46. This performance wouldn’t be met or surpassed until Google’s IPO in 2004.

The Collapse

By Mneilly — Excel charts created using data directly from SEC 10-K reports for TMTA.

Unfortunately, it was all downhill for Transmeta. While Sony, Fujitsu, Sharp, and others did release Crusoe-based chips, right before Transmeta’s IPO IBM canceled their much vaunted Cursoe-based ThinkPad. Additionally a deal with Compaq failed at the last minute.

Three weeks after Transmeta’s IPO NEC announced a recall of 300 Crusoe powered chips, and Sony had to announce shortly after NEC that their Crusoe powered laptops might also be affected.

To continue to pile on to Transmeta’s woes their transition from a 180-nanometer process to a 130-nanometer chip was delayed forcing Transmeta to announce they couldn’t guarantee enough chips could be produced for their vendors. This naturally caused PC vendors such as Toshiba to ditch the Crusoe due to inability to purchase chips.

In March of 2001, David Ditzel stepped down as CEO and was replaced with Mark K. Allen. During his short reign, two more founders left the company.

Then sevens months later in October of 2001, they replaced Mark Allen with Murray A. Goldmann, and by January of 2002, their stock was trading in 1–3 dollar range down over 90% from its peak.

At the same time, Intel did not take this threat laying down, quickly releasing a low-power Pentium III chip and then in 2003 launching their Pentium M processor, Intel took the wind out of Transmeta’s sails.

Finally, Transmeta was the last uber-stock of the early-21st-century dot-com bubble which was well on it’s way to popping by Transmeta’s IPO, giving them a tiny runway for mistakes. Transmeta couldn’t handle the onslaught. In July of 2002, they had their first round of layoffs cutting their staff by 40%. Then on June 17, 2003, Linus Torvalds left Transmeta to work full-time on the Linux kernel.

On October 14, 2003, Transmeta announced their Efficeon processor which claimed to have 2x the performance of the original Crusoe, but it was too little too late.

In January of 2005, Transmeta announced that it was moving away from being a semiconductor company and was going to focus on licensing intellectual property. In March of that year, they announced another layoff of 68 people. In May of 2005, Transmeta announced it was going to sell the Crusoe line to Hong Kong’s Culturecom Technology for $15 million in cash. Ultimately export regulation killed the deal, and it fell apart in February of 2006. In August of 2005 they announced their first-ever profitable quarter, earning $10.1 million.

Transmeta continued limping along until October 2006 when it sued Intel for violating their patents in every processor since the Pentium Pro. But at this stage, Transmeta was a shell of a company. In February of 2007, they shut down their Engineering division, announcing they would no longer develop or sell any hardware focusing only on the development and licensing of their IP. On October 24, 2007, Transmeta settled with Intel for 250 million dollars, but Transmeta had to agree never to manufacture x86 compatible chips ever again, though at this point it’s unlikely Transmeta was in a position to produce anything but debt.

In November of 2008, Transmeta announced they were going to be acquired by Novafora, and the deal was finalized on January 28, 2009.

Transmeta which started out with such a bang, ended in a quiet whimper.

What happened to Transmeta?

So what doomed Transmeta?Here are three significant reasons why the company ultimately collapsed:

Economic Recession

NASDAQ Composite Index

No business gets to pick the economic environment they operate in, but the one Transmeta jumped into was well on its way toward the Dot Com Bust, and the early 2000s recession. In March 2000, it was announced that Japan had once again entered a recession, this was particularly bad news for Transmeta as Japan had proven to be a strong market for the small, battery-life critical devices Transmeta had in mind. But that same year Japan hit a then-record 5% unemployment rate, putting a significant drag on consumer spending.

The US was right behind Japan in the recession race. Only 6 months before Transmeta had their IPO, NASDAQ had started it’s long slide down. Two days after their IPO, Pets.com,one of the most hyped companies of the era, went out of business. Before the recession was over NASDAQ would lose 78% of its value.

In short, January 2000 was probably one of the worse times in history to launch a new, capital-heavy product in a very conservative market.

Intel’s Response

“Transmeta is not relevant,” Jerry Sanders, CEO of rival Advanced Micro Devices, asserted in a recent conference call. “Intel has effectively kept them out of the marketplace.” — (Transmeta : Are the chips down? : Tech News : CNET.com)

Intel did not take kindly to this upstart chip maker moving into their x86 turf, particularly since at the time they were under pressure from AMD Athlon chip. They immediately kicked off an energy-efficient chip project, launching the Pentium M in 2003. Intel also leaned on their relationships with PC Vendors and offered lower prices if companies only used Intel products in their lineup. Simultaneously, Intel slashed prices on its laptop processors, pushing Transmeta down from the standard laptop line to the much less popular “mini-notebook” line. Throughout its entire existence, Intel was able to exert a tremendous amount of downward pressure on Transmeta.

Failure to Execute
Transmeta had a very narrow path to success and had to execute almost flawlessly to pull it off.

They did not.

Right out of the gate, Transmeta proved they were not ready to perform. They announced the Crusoe in January of 2000 and had no product ready for market for another 10 months. Also their early products immediately had recalls, which did not build consumer confidence.

Transmeta also took a beating from the tech press because, for the most part, the Crusoe wasn’t being used in the best products. A typical review was like this one from November 2002 from cnet:

“In application performance, the Compaq TC1000’s Transmeta Crusoe TM5800 processor crippled the system. While this processor clocks in at 1GHz, it still can’t compare to the Pentium III-M-800MHz processor of the Fujitsu Stylistic ST4000 and especially not the 1.3GHz Pentium III-M processor of the Toshiba Portégé 3505. Indeed, the Compaq’s Crusoe processor turned in the worst times of all the tablets we tested.”

The reviews would normally then go on to praise Crusoe’s battery life, but it was too little too late.

Additionally, Transmeta failed to transition from 180-nanometer fabrication to 130 successfully. During critical early years, they couldn’t make chips to sell. As a chip vendor, that’s a pretty unsustainable place to be and sours relationships with companies you depend on to sell your product.

Transmeta failed to beat Intel in speed (or at least speed perception, and to consumers perception is reality), quality (how good can something be when it has a recall as soon as it launches?), and time to market(Intel had already promoted a new life of energy-efficient processors by the time Crusoe shipped.) From a consumer perspective the product was DOA.

But, perhaps Transmeta could have turned things around if they sold a lot of cheap devices and build some brand recognition, but they failed at that too. By constricting supply at the very time they should be moving the most of their product, and building those relationships.

The story of Transmeta is the story of a great idea with poor execution. By coming out so late, they missed the tech bubble of the 90s that would have given them some buffer. And yet, even though it took 5 years to make, the chip and it’s rollout still seemed rushed and unprepared. Ultimately, that was all it took, and despite having genuinely outstanding technology, it could never make the impact it needed to survive.

--

--

Madmedic

I knew who I was this morning but I’ve changed a few times since then.