We live in a DeepSeek era, when every bit of scientific progress is market-moving, however great or small. 

Case in point, Microsoft’s announcement Wednesday of a “topological qubit” for quantum computing. Journalists such as The New York Times’s Cade Metzhave stumbled on this as if is brand-new, but it’s not. 

Quantum stocks surged on this bit of progress Wednesday, after having had a rough start to the year. (The stocks are overpriced, so they need headlines like this to keep the shares aloft.)

The immediate news event is a highly technical report in this week’s Nature Magazine by Microsoft’s Morteza Aghaee and dozens of colleagues describing a kind of chip for manipulating quantum dots and nanowires. The more easily digestible read is a Microsoft blog post describing the chip as capable someday of holding millions of topological qubits that will work in concert to do practical computing. 

It was a tough set-up for Arista Networks going into Tuesday night’s earnings report, and the eight-percent drop on Wednesday reflects that hurdle.

It was a good report, the company beating with reported results, and with its forecast for this quarter’s revenue. But the sole focus had been on one number, $750 million, which is the number that CEO Jayshree Ullal had projected all the way back in November of 2023 as the likely revenue from artificial intelligence this year.

Even though Ullal reiterated that expectation Tuesday, what people really had wanted was an increase, which they didn’t get.

Shares of Intel are up ten percent Tuesday, adding to several large jumps last week, bringing it to a twenty-one percent rise in the past thirty days, on speculation the company’s manufacturing operations might be a buyout candidate.

There’s considerable skepticism about it all on the Street, considering what a complex situation Intel has become: planned spin of manufacturing already having been announced, engagement with the U.S. CHIPS Act for funding of factories, etc.

The rise in Intel shares came to a head on Friday, as Bloomberg’s Mackenzie Hawkins and Ian King reported that Taiwan Semiconductor Manufacturing, the world’s largest chip manufacturing arm, is in the “early stages” of talks to take a controlling stake in the manufacturing unit, at the request of U.S. president Donald Trump, citing a single unnamed source. 

• DeepSeek was all over the papers weeks ago, January 27th. Well, how quickly we forget. The Chinese AI model isn’t coming up as much in the news now, and in the two weeks since it crushed tech stocks, from January 27th through February 14th, the Nasdaq Composite Index has risen by 3.5%. 

• If you bought that day any of the TL20 stocks worth considering, you probably got a pretty good return. The TL 20 group of stocks worth considering is up 9.4% since January 27th.  As for DeepSeek, what seemed to be a dramatic breakthrough on Monday, January 27th, has since come to be seen as kind of more of an evolutionary development of AI. 

• The price change for Nvidia stock from January 27th through February 14th is a nifty 17% return; if you held your shares, you made back almost all your money. 

Within the software market, you may or may not know that there’s very vibrant competition in the category of financial software. That includes software for “human capitalmanagement,” such as HR functions, payroll, and for functions of the finance office such as general ledger and other things needed to close the books.

I’ve outlined both the small names and the giants in the accompanying table, which shows how they did in this earnings reporting season.

This is a group that has not been much in favor, within an overall very tough market for software stocks (for anyone other than Palantir or Monday dot com.) The average gain for the group over the past twelve months is just sixteen percent versus twenty-seven percent for the Nasdaq Composite in that time.

Thursday evening was a disappointment for software maker Twilio following an auspicious “analyst day” event last month, the shares sinking fifteen percent to $125.78 as the outlook for this quarter missed expectations for both revenue and profit.

Most analysts are taking the drop in stride, as the shares had surged in the past three months by thirty percent. The disappointment is a case of “heightened expectations,” they figure. The main point the bulls emphasize is that the forecast for this year offered at the January event, for seven to eight percent revenue growth, and a big leap in profit, was reiterated Thursday by CEO Khozema Shipchandler and team.

Most price targets are increasing this morning, as analysts take the shortfall in stride in the context of the bigger picture. “The tone of the call was constructive,” writes Brian White of Monness, Crespi, who reiterates his Buy rating and raises his  target to $165 from $152. Shipchandler and team were “optimistic about 2025 with management highlighting a positive business environment.”

“As we move into the AI world, the data center architectures are changing quite significantly … We also need to evolve to meet those changing architectural needs.”

Wednesday evening’s earnings featured some reversals of fortune.

Reddit, a darling of momentum investors, dropped eleven percent as it seemed to fall prey to Google’s whims. Cisco Systems, which has been rather dull of late, surged six percent as the company got some credit for its artificial intelligence luster.

Other things remain the same: The Trade Desk, which has a very checkered history of ups and downs on its earnings reports, plunged twenty-nine percent as its revenue outlook missed expectations. And AppLovin, the video-game marketing tools company that is also a momentum darling like Reddit, continued its winning streak, its shares surging twenty-eight percent in late trading, adding to a seventeen percent gain since the start of the year.

Tuesday night saw the third report as a public company of software maker OneStream Software since its July initial public offering, and it wasn’t pretty, with the stock selling off by almost eighteen percent in late trading.

At the after-hours price of $24 or so, the stock is below its first-day close at the IPO of $26.85.

The question is how much of the “uncertainty” the company cited in its outlook is indicative of the broader software market.

OneStream, a twelve-year-old company based in Birmingham, Michigan, aims to reinvent the office of the chief financial officer with its programs. The results were fine, but the forecast for this quarter came in below consensus. That’s just not okay considering the stock was pricey, with a multiple of forward sales of thirteen times.

Steady-as-it-goes is the picture on Monday for Monday dot com, the small-cap (thirteen billion dollars in enterprise value) workgroup software maker that is a favorite of the growth investing set.

The company once again turned in results and an outlook that topped analysts’ expectations, but this time, the shares are up over twenty-six percent versus the November report, when results were also good but the stock sold off by fifteen percent.

This is a common pattern with Monday, this flip-flopping in stock performance from one report to another. A lot has to do with very short-term expectations. William Blair’s Arjun Bhatia today writes that there had been a a lot of “noise” in the two months preceding this report, with investors modeling lower and lower growth for 2025.

One of the biggest winners of earnings this season is American Superconductor, rising thirty-four percent after Wednesday evening’s results beat the estimates of the three analysts following the company, as did its outlook for this quarter’s revenue.

Take the jump with a grain of salt: The stock has surged like this in past. Still, the trend is a very good one. It’s the eighth quarter in a row the company’s results and outlook have surpassed expectations, and earnings power is a bright spot.

The Street will be most focused on encouraging words from CEO Daniel Patrick McGahn that hint at AI data center opportunities.

Being a second fiddle to Nvidia continues to dog Advanced Micro Devices’s stock, despite pretty stellar results. 

The company Tuesday evening topped analysts’ expectations for the December quarter, and with this quarter’s outlook, but saw its stock sell off by ten percent in late trading.

This was a repeat of October’s report, when the numbers were strong but all anyone cared about was seeing higher and higher AI chip revenue. 

The problem has been rising expectations that keep running ahead of even very good results. Although the company turned in five billion in AI-related chip sales for all of 2024, just as CEO Lisa Su had promised in October, the December quarter’s AI number, $1.75 billion, was below consensus for $1.95 billion. 

Shares of Palantir are up over twenty percent following Monday evening’s better-than-expected results and outlook. As I noted yesterday, it’s one of the most expensive stocks ever, trading at a hundred and thirty-eight times this year’s projected profit (Ebitda).

There are different ways think about that valuation. One way is resignation, from the bears: It’s too high but it’s probably going to go even higher.

That’s the case for Morgan Stanley’s Sanjit Singh, who raised his rating to Equal Weight from Underweight, and hiked his price target to $95 from $60 — which is still below today’s $102.75. He just can’t find “a clear downside catalyst despite an expensive valuation.”

Shares of software maker Palantir are getting even more ridiculously overpriced this evening.

The stock is up twenty-three percent in late trading Monday, at $102.52, after another beat-and-raise earnings report. 

With two hundred and six million shares outstanding, that would give the company an enterprise value of $206 million, and equate to a multiple of projected sales, based on what will be the new consensus tomorrow, probably, of, say, $3.8 billion in sales, of sixty-nine times.

Palantir was already ridiculous, but now it’s really ridiculous.

The last time anything was worth that much money was in 2020, when Snowflake came public, and was for the a time the most expensive stock anyone had ever seen, in any industry. 

The numbers show strong upside. 

It’s not as bad as last Monday’s DeepSeek plunge, but leading tech names are down, including Tesla, off six percent, Apple down four percent, and Nvidia down almost four percent, after the White House’s announcement Saturday of tariffs imposed on goods from Mexico, Canada and China, the U.S.’s three largest trading partners.

Despite everyone knowing this was coming, there’s disagreement and a lot of uncertainty about how it hits tech.

The issues may loom larger for groups such as discretionary consumer goods, but rising prices of electronics, and electronic components, has clearly spooked the entire tech group.

• The massive sell-off in the Nasdaq Composite Index on Monday the 27th from the Chinese AI model DeepSeek had the result of driving up some shares dramatically amongst the biggest tech names by week end.

• From Friday the 24th to the subsequent Friday, the 31st, which captures that big plunge on Monday, stocks fell hard; but, Monday to Friday, you get a big gain for the week.

• Nvidia was the biggest decline, of course, from Friday through Friday, down by 16%. But from Monday through Friday, it was actually up over 1%.

• Arista Networks, from Friday to Friday, was down 11%, but from Monday's close to Friday, was up 15%.

• A lot of people swooped in to get deals after Monday. The bad news: A lot of things were more expensive at week’s end.

We’re in the thick of earnings season, and here’s a rundown of what we learned the last twenty-four hours.

SOME BRIGHT SPOTS IN SOFTWARE

In the struggle for the software group to bounce back, Thursday brought some bright spots.

“DevOps” software maker Atlassian is a come-back story that appears to be going in the right direction. The stock was a dog the past twelve months, up only four percent. Shares are up twenty percent in late trading after the company beat expectations for the quarter and for this quarter’s outlook, and also raised the revenue outlook for this fiscal year for the second time in a row. The original growth outlook in August was just sixteen percent; it’s now as much as much as nineteen percent. What’s helping is a price increase just recently unveiled. Investors are also responding to higher-than-expected gross profit margin.

If you’re still shaking from DeepSeek’s disrupting of AI stocks, take heart: Microsoft and Meta Platforms told analysts they’ll spend even more this year than last — a lot more — driven not entirely by artificial intelligence, but with no small emphasis on that.

Management at both companies indicated that in one way or another, economic gains of the kind that DeepSeek represents ultimately lead to more computer use, broadly speaking, not less. And they’re spending with that expectation.

The numbers for these companies’ spending is rather staggering.

Microsoft CFO Amy Hood told analysts the company will increase capital spending from $55.7 billion last year to rise  probably over fifty percent, to $84 billion, based on the rate of spend in the first two quarters of the year.

Although it seems DeepSeek is all anyone can talk about, we are actually in the thick of earnings season, lead this morning by upbeat reports from chip-equipment giant ASML and fiber-optics giant Corning. 

ASML shares are up almost four percent after the company reported record revenue, and forecast this quarter’s revenue well above consensus, and the outlook for the full year 2025 is in line with Street consensus. (Note that reports of last the headline number vary: Thomson-Reuters has the company missing expectations, while FactSet lists it as a beat.)

The funny thing is that the stock rise comes as ASML’s CEO, Christophe Fouquet, and CFO, Roger Dassen, told the analysts that next year, they’re going to give less information. 

Remember that ASML’s report in October was a big disappointment relative to expectations that the duo had set up in July. I noted at the time that there’s something of a credibility gap with such sudden surprises. 

It’s day two of unpacking the meme of DeepSeek AI, the Chinese artificial intelligence startup whose perceived revolution in processing trashed the market on Monday.

There’s a lot of verbiage coming fast and furious from Street observers this morning, of all different sorts.

It’s frankly stunning how many parties have glommed onto the DeepSeek thing to comment.

Morningstar’s lead asset finance expert, Victor Leung, for example, speculates as to whether the DeepSeek risk to AI funding could imperil credit ratings of data center companies. (He actually concludes that cheaper AI “could be beneficial to the credit underpinnings of the entire AI data centre ecosystem in the long run.”)

As for the pummeled stocks, they are nicely recovering today, with Nvidia up over six percent following Monday’s seventeen percent decline, Arista Networks up over two percent, Astera Labs up almost four percent, Ciena up five percent, and Taiwan Semiconductor up over three percent, among others.

• Special Stargate DeepSeek Edition. The week that ended January 24th was the week of the inauguration of U.S. President Donald Trump. Trump had a photo op in the White House with a group of tech individuals led by Masayoshi Son of Japanese conglomerate SoftBank. They promised what seemed to be a bonanza in artificial intelligence funding, half a trillion dollars over several years, specifically for AI data centers. That boosted already elevated AI-related stocks, including NVIDIA.

• Then, Monday, January 27th, stocks plunged as a result of a fervor over DeepSeek, an artificial intelligence startup operation created in China by a hedge fund entrepreneur.

• You have the intersection of two hyped events, Stargate and then DeepSeek. Tthey're both overhyped developments.

• Masayoshi Son has a history of making grand pronouncements that have less to them than would initially appear. Monday's deep seek announcement was a sort of reverse-hype, picking out one technical development by DeepSeek and making it the answer to all of the hype about GPU chips. Two hyped events perfectly connected.

Tech shares are falling out of bed Monday, and I had to quickly rifle through my memory for the name DeepSeek.

Shares of Nvidia are down over twelve percent, and some AI-linked names are also down, most notably shares of fiber-optics maker Coherent, down eighteen percent, AI networking firm Arista, down sixteen percent, and Nvidia competitor Broadcom, off twelve percent.

DeepSeek, the creation of a Chinese hedge fund, as impressive as it is, is probably not the end of tons of spending on Nvidia GPU chips, and all the rest. But it does show how the AI hype can burst when another idea becomes suddenly viral.

What’s got everyone excited is that DeepSeek has detailed spending ten to twenty times less on Nvidia chips to run its “DeepSeek V3” than top-of-the-line programs such as OpenAI’s GPT-4o, and yet can match or exceed 4o in capability.

“I think more than anything what investors want from us is an innovative company that's really well run. And I think that we are nailing the really well run part of that.”

The Street is fairly giddy today about the implications of “Stargate,” the corporation that U.S. president Donald Trump unveiled Tuesday with the support of Japan’s SoftBank conglomerate, Oracle, OpenAI, and cooperation from Arm, Microsoft and Nvidia.

The focus is an initial hundred-billion-dollar investment, to be somehow obtained by SoftBank, and eventually as much as five times that, to be spent on what Trump refers to as “colossal” data centers. These are data centers specifically for artificial intelligence, initially in Abilene, Texas, according to The Financial Times’s George Hammond, Myles McCormick, and David Koehane, “construction of which is already under way, according to the companies — before expanding into other states.”

This is, overall, nothing new, unless you’ve been hiding under a rock for many years. The last thing in the world that needed any help, from any government, was massive spending on AI, which has already been just about the only thing in the tech landscape that is a slam dunk.

Everything that was concerning about Netflix’s third-quarter report in October was forgiven Tuesday night, as the company crushed subscriber expectations, and raised its revenue forecast and operating profit forecast for 2025. Shares surged after hours and are up thirteen percent as I write this. 

The question now for investors is whether Netflix should be among the most expensive of Mega Tech stocks, which it currently is at forward multiple of about thirty times. 

The results, first of all removed for most investors, the concern that had cropped in October that Netflix might be “stalled.” The company said that it added nineteen million subscribers, the highest in its history, and double what the Street expected, and it said its revenue for this year will be half a billion dollars, and also raised its profit margin outlook by a point to twenty-nine percent. 

• The week that ended January 17th was a favorable one for stocks, bouncing back from a sell-off the week before. The Nasdaq Composite Index closed the week up 2%, the Standard and Poor’s 500, up even better, 3%, and the TL20 group of twenty great companies whose shares are worth considering was up the same as the Nasdaq, 2%. 

• What did well: quantum computing names continue to soar, with Rigetti Computing up 62%, Quantum Corporation up 56%, Quantum Computing up 51%, and IonQ up 40%. That’s on top of some amazing gains in several of these, with Rigetti in the last twelve months up 873%. 

• Wolfspeed, the leader in silicon carbide chips, up 28% percent for the week on a feeling, a hope, that maybe the SIC market is coming back with some return to health of some the EV automotive market. 

It’s been a long time since I can remember anyone making a case for investing in Sony with any great enthusiasm, so, I was intrigued when I noticed on Thursday Bernstein’s David Dai initiating coverage of the stock with an Outperform rating, and a price target of ¥3,900 Japanese Yen, which would be about twenty-two percent from Thursday’s price on the Tokyo exchange of ¥3,100.

The shares just edged out the Tokyo Topix index in the past year, and the three-year return was miserable, a little under five percent versus thirteen percent for the index.

But, now, Dai sees better times ahead thanks to profit growing faster than revenue for the first time in a while. “In contrast with an indifferent 2022-2024, we expect a Sony more focused on efficiency to deliver earnings surprises,” writes Dai.

Taiwan Semiconductor Manufacturing, the world’s biggest contract chip maker, is the main supplier to Nvidia, and, when it comes to spreading the ebullience about all things artificial intelligence, Taiwan Semi is Nvidia’s partner in crime.

Almost every quarter, the remarks from TSM’s CEO, C.C. Wei, are the soundbite that backs up Nvidia’s CEO, Jensen Huang. Back in October, Wei spoke of “insane” demand for AI chips. For the latest earnings report, on Thursday morning, Wei didn’t need to use any hyperbole: The number speak for themselves.

AI-chip-related revenue was a “mid-teens” percentage of all of Taiwan Semi’s revenue in 2024, noted Wei, as that revenue tripled, year over year, driving Taiwan Semi’s total revenue for the year up a very healthy thirty percent. But, this year, he said, AI-related revenue will double again. Clearly, that means AI-related revenue is going to be even higher this year as a percentage of Taiwan Semi’s business.

I noted Tuesday that expectations for Microsoft’s “co-pilots” have met with disappointment, as folks such as Morgan Stanley’s Keith Weiss have had to revise down their revenue forecasts amidst lackluster demand for the stuff.

But hope springs eternal, and the enthusiasm that once attended co-pilots has now attached to “AI agents.” Some observers are starting to dream big, big numbers, just as they once did for co-pilots.

Meta Platforms is Exhibit A in that respect.

Meta is probably one of the companies that has had a better reception on the Street as far as AI usage, and the company is getting the benefit of the doubt it will make a lot of money off of AI agents.

As you may have heard, the White House on Monday issued a decree by outgoing president Biden placing restrictions on which countries can have access to U.S. artificial intelligence chips, which it claims was necessary to “ensure” that “U.S. technology undergirds global AI use and that adversaries cannot easily abuse advanced AI.” 

The announcement is being referred to as the “AI diffusion rule.

There’s been very little response, probably for exactly the reason I mentioned on Friday, which is that the Street sees little prospect that such a measure will be upheld by the incoming administration of Donald Trump.