Nvidia’s earnings are no elixir to reverse the market’s sudden momentum breakdown

Eos Energy Enterprises is surging in early trading after announcing the official start of commercial production at its second automated battery manufacturing line.

In a statement, the company said this milestone positions it to scale production of its proprietary zinc-based long-duration energy storage systems to meet rising commercial demand.

Management touted the enhanced efficiency of this facility, with design upgrades slashing raw material travel by 86% and shortening the physical production line length by 40% compared to Line 1.

“Battery Line 2 demonstrates our ability to continuously improve as we scale,” said John Mahaz, Chief Operating Officer of Eos. “It validates that our manufacturing system can be replicated and scaled with discipline.”

The battery energy storage company confirmed that while subassemblies will continue coming online through the early third quarter, full production capacity is targeted for the fourth quarter of 2026. The ultimate goal is to hit an aggregate 4 gigawatt-hours of annual manufacturing capacity by the end of 2026. Management also highlighted that Battery Line 1 already surpassed its full-year 2025 output within the first 164 days of 2026.

Today’s announcement builds on recent operational momentum for Eos, which posted better-than-expected Q1 sales and announced a joint venture with Cerberus Capital Management in May. However, shares are still down 37% year to date.

For the full year, Eos still expects to achieve revenues between $300 million and $400 million, in line with its previously provided guidance.

Qualcomm is in talks to acquire AI chip design firm Tenstorrent for $8 billion to $10 billion, according to The Information.

This transaction, if completed, would be another concrete signal of the San Diego-based chip company’s attempt to carve out a niche in the upstream AI space (data centers), rather than focusing on end-user devices.

Qualcomm’s key business of handset chips has fallen on hard times, particularly in China, due to the memory chip shortage.

Less than eight weeks ago, the chip company was the lowlight in the Philadelphia Semiconductor Index, down about 20% year to date.

Shares proceeded to surge over 60%, buoyed by optimism that the rising AI tide will lift all boats. With the release of Q2 earnings, CEO Cristiano Amon said that initial shipments of AI chips to a “leading hyperscaler” were on track for later this year, and to expect more on the company’s AI growth plans at its investor day on June 24 (next week). Last month, Bloomberg reported that Qualcomm is poised to sell “millions” of AI chips to TikTok parent ByteDance.

Established AI chip giants and hyperscalers alike have reached agreements with or gobbled up burgeoning AI chip companies as the boom rolls on. In December, Nvidia announced a major licensing deal with AI inference specialist Groq, while Meta bought AI chip startup Rivos in September.

A major theme of this year is that American companies are once again becoming major sellers of stocks.

For years, companies did the exact opposite: buying back trillions of dollars’ worth of shares, a practice that juiced earnings and was seen as a safe option for management teams that had run out of good-enough projects to allocate their capital to. Just look at Google, which is wiping out more than two years’ worth of buybacks with an $85 billion offering, while Meta reportedly mulls an equity raise of its own.

Now, the mantra is that investment opportunities in AI — particularly as suppliers to the arms race — are a source of future returns that are also key to sustaining higher growth. In short, capex is king, and buybacks are admitting that you don’t have enough investment opportunities that allow you to benefit from the AI boom. Raise debt, raise equity, raise anything — just make sure you’re spending, and the market will reward you. A Goldman Sachs basket of companies with elevated capex relative to peers is besting stocks with the strongest buyback yields by some 30% — the most ever.

This is leading to some major divergences in accrual-based profit measures, like net income and free cash flow (which takes capex into account), for companies like Oracle.

Of course, the rest of the AI complex doesn’t care whether the cash spent on the next data center was raised via debt or equity. More funding for the AI build-out is more funding for the AI build-out. Indeed, if we took capex to a bazillion dollars, that spending would still be accretive for aggregate earnings in the first year (assuming all the recipients of the capex binge were public stocks). Yes, eventually the depreciation on those assets starts to be felt and we’d normalize lower, but in the short term, it’s a boon to the stock market’s bottom line.

This is why Oracle’s chart is actually just a more extreme version of the wider market; free cash flow used to be about 90% of aggregate net income, and now it’s hovering around 75%, per estimates compiled by Bloomberg.