Tower Semiconductor (NASDAQ: TSEM) reported another record quarter, delivering $460.1 million in second-quarter revenue, up 24% year over year, alongside $0.79 in diluted earnings per share. The company didn’t stop there.
It guided third-quarter revenue to a record $520 million, raised its 2028 revenue target to $3.6 billion, and projected $1.2 billion in annual net profit by the end of the decade. Though the shares opened down 3.43% to $231.31 anyway, the selloff wasn’t driven by weak execution.
Tower produced record gross profit, record operating profit and record net profit while lifting gross margin to 30%, operating margin to 20%, and net margin to 20%. Those gains arrived as the company continued pouring hundreds of millions of dollars into new manufacturing capacity rather than harvesting cash from existing assets. This is where you need to pay attention because these earnings weren’t just about another strong quarter. They revealed a company changing one of the oldest assumptions in the semiconductor industry: how chip factories get built.
Every Chip Factory Starts With A Gamble, Tower’s Doesn’t.
Semiconductor manufacturing has always been a capital-intensive bet. A company commits billions of dollars to build or expand a fabrication plant years before knowing whether demand will still be there when the first wafers roll off the production line. If customers disappear, the industry is left with underutilized factories, shrinking margins and painful write-downs. Every major downturn in the semiconductor cycle has followed some version of that script.
Tower Semiconductor (NASDAQ: TSEM) is flipping that on its head. The company disclosed that it received $290 million in customer prepayments earlier this year, largely tied to 2027 capacity reservations. During the earnings call, management also revealed that it has already secured approximately $1.3 billion of silicon photonics revenue commitments for 2027. Then came the most revealing remark of the call: the additional manufacturing capacity now being built has already been requested by customers and is “spoken for.”
Rather than expanding first and hoping customers eventually fill the factory, Tower is securing long-term demand before pouring concrete and installing equipment. The company is still committing heavily to growth – a $920 million expansion program remains underway – but a meaningful portion of the commercial risk has already shifted. Put another way, customers are reserving tomorrow’s production years before those wafers exist, giving management a level of visibility that is unusually rare in an industry better known for boom-and-bust cycles.
The Game Of Specialization
Tower’s expansion could easily be mistaken for another beneficiary of the semiconductor recovery. The timing certainly fits. For nearly two years, chipmakers wrestled with excess inventories after the pandemic boom faded, leaving factories underutilized and customers working through existing stock rather than placing new orders. That cycle has started to turn as AI infrastructure spending remains robust and industrial demand gradually recovers. Tower, however, is riding a much more specific wave: silicon photonics, a technology becoming increasingly important as AI systems demand faster, more energy-efficient data movement between processors.
And the numbers show how fast that business is scaling. Silicon photonics revenue climbed more than 270% year over year and 60% sequentially, reaching an annualized run rate above $680 million. Management expects that figure to exceed $1 billion by the fourth quarter as new capacity comes online.
That helps explain why Tower raised its 2028 revenue target to $3.6 billion from $2.84 billion, while projecting gross margins to expand from 39% to 45%. Indicating that the company is now expanding into products customers are already lining up to buy, allowing each new factory to generate more profit than the last.
More Like Consolidation Than A Change In Trend
Tower’s rally into earnings explains much of the post-results pullback. After rebounding from July’s low near $190, the stock climbed roughly 30% before stalling at its 50-day moving average of $247.52, a level that has capped every rebound since the May peak near $320. When the earnings report arrived, traders who bought the anticipation locked in profits, sending the shares down 3.43% to $231.31, almost exactly on the 20-day moving average. (chart)
The broader trend, however, remains constructive. The stock continues to trade well above its 200-day moving average at $170.39, while July’s $200 support remains intact. The market isn’t necessarily rejecting Tower’s expansion plans here. It’s asking management to prove that customer prepayments, long-term capacity commitments and an ambitious 2028 roadmap can translate into sustained earnings growth. If that evidence continues arriving each quarter, this chart suggests this pullback may look more like consolidation than a change in trend.
Contracted Growth Is Worth Buying On Weakness
Tower’s quarter showed more than record revenue and record margins. It showed a business expanding against demand that has already been committed. The company guided third-quarter revenue to another record $520 million, raised its 2028 financial model to $3.6 billion in revenue and $1.2 billion in net profit, while customers continue reserving future production years in advance.
But the risks haven’t disappeared. Tower still needs to execute a $920 million capacity expansion, keep wafer pricing firm and bring new fabs online on schedule. Those are execution risks, not demand risks. For investors, that’s important to keep in mind as the stock may need time to work through its post-earnings consolidation, while the business continues moving in the opposite direction.
