Taking Stock / Tower of Jelly

"What Idan Ofer did was despicable and stupid."

Carmel Vernia is boiling mad and he wants revenge. Two weeks ago the departing Tower Semiconductors CEO and chairman received an unusual review from the chairman of the The Israel Corporation, the controlling shareholder in Tower. In a truly rare move, Ofer pointed the finger of blame at Vernia: "There was a management problem at Tower," he accused, adding that the company had failed to fulfill its business plans.

Vernia decided to shoot back and in an interview with Haaretz, he shed new light on one of the biggest investments made in Israel for decades: Fab 2, Tower's second chips foundry in Migdal Ha'emek, a venture in which the state, the banks Leumi and Hapoalim, have invested $750 million and counting.

1. "That is not entirely wrong," replies Vernia to our question, whether, perhaps, a facility like Tower had no right to exist in Israel.

We have asked that question time and again for years, in conversation with people at The Israel Corporation and Tower. The answer was different this time.

Tower's team lacked the experience, the ability and the knowhow in chipmaking to make a go of it, Vernia alleges. "Tower relied on the team and management of Fab 1, but their knowhow was irrelevant to the new fab, which was supposed to compete with world giants. We kept having childhood diseases and paid heavily for the lessons learned," he says.

2. So why did Tower set out to build Fab 2, anyway? Vernia offers an original suggestion: The former CEO, Yoav Nissan-Cohen, "was a talented, charismatic guy who managed to persuade a lot of people.

"Nasdaq was at 5,000 points and the biggest chipmaker, TSMC, was earning 50 percent of its turnover," Vernia explains. That was enough to convince Idan Ofer, the state and the banks to launch the $1.5 billion project.

3. "Amateurish and negligent," Vernia describes the management of Fab 2 from day one. He describes problems in financing (not settled in advance), and in operations (orders were not secured in advance). Also, the business model of bringing in strategic partners who would also be major clients backfired, because when the market tumbled, only one of the partners - Sandisk - continued to place orders.

4. "Idan Ofer stopped coming to meetings of the management committee nine months ago," Vernia says, hence his astonishment at The Israel Corporation chairman's statements. "Maybe when he smelled failure in the air, he figured, let me replace myself on the management board with Yossi Rosen" - the general manager at The Israel Corporation - "cast all the blame on Vernia, and find another bandwagon to ride."

5. Vernia reverses that finger of blame back at Ofer, but also admits that he himself wasn't the most suitable person for Tower.

"I'm not suited to manage a chip manufacturer," he admits somewhat belatedly, revealing that he'd been named after the board spent a year and a half trying to find somebody more appropriate. Tower's inability to find a manager was another flop of the management board, or proof that the company had no right to exist in the first place: no manager of repute in the chips industry wanted any part of it.

6. The fiasco cannot be dismissed as a matter of timing. The same year Tower decided to build the fab, a similar one was built in China, called SMIC.

The Chinese management did a few things Tower's didn't: They assured big orders at rock-bottom prices, allowing the facility to operate at full capacity from day one; and they finalized financing in advance. The result is that SMIC is selling at a billion dollars a year, eight times the turnover at Tower, and is trading at a market cap of $4 billion, which is 40 times Tower's.

7. It was already clear a few years ago, alleges Vernia, that the probability of returning investment in Tower was small. Prices in the chips industry were plummeting, and every delay in finishing the fab was reducing the chance of returning the investment. Manufacturing facilities like Tower must be completed in 18 months to remain technologically relevant, while in Tower's case it has taken seven years.

8. To contend with the plunge in prices, Tower devised a strategy of developing niche products: additions to chips such as embedded flash or image sensors.

"That was the right direction," Vernia avers, but says Tower didn't require a fab like it planned to build to achieve it. "Instead of pouring $1.5 billion into a new fab, you could buy a used one for $100 million or $200 million," he explains. "You certainly don't need a new fab to make niche products."

Vernia may be skewing some things to his own convenience. Nonetheless, clearly the decisions Ofer and then Vernia made to expose all this dirty laundry is educational for investors, as to how things were run at Tower Semiconductor, and what its chances are.

The flying mud may present an opportunity for the state, the banks and shareholders to digest that huge companies can make huge mistakes, and that what seems to make very little sense usually does make very little sense.