
Sep 3, 2026 · 28 min
Listen from 3:59
Listen at 3:59
Data center backlash tests AI infrastructure stocks
Is the Data Center Investing Trend in Trouble?
The episode connects local resistance and physical constraints around data centers to the growth assumptions embedded in several AI infrastructure investments.
- 1Snowflake’s AI workloads and consumption model support growth, but valuation and margin risks remain significant.
- 2Data centers can bring jobs and development while intensifying disputes over electricity, infrastructure, and local control.
- 3A potential construction slowdown could pressure infrastructure stocks, making current investment cases more important than past purchase prices.
Don't miss
The hosts connect a possible data center construction pause to the outlook for Marvell, Celestica, and Sterling Infrastructure.
The brief
Jon Quas, Lou Whiteman, and their co-hosts begin with Snowflake’s strong quarter, examining revenue growth, guidance, margins, customer spending, and the AI workloads driving demand.
Snowflake’s consumption-based model and Cocoa coding agent could extend its AI opportunity, but the hosts balance that promise against valuation and margin risks.
The conversation shifts from software to real-world bottlenecks as moratoriums and public opposition in New York and Texas challenge rapid data center construction.
Data centers may create construction activity, permanent jobs, and local development, yet those benefits compete with rising electricity demand and strained infrastructure.
A pause in construction could affect Marvell, Celestica, and Sterling Infrastructure, with power, labor, chips, equipment, capacity, and capital shaping the outlook.
The mailbag closes on a broader investing discipline: judge a losing position by its current case and price, not by the purchase price already paid.
What was said on this episode
39 statements · 17 positive · 17 negative · 2 mixed · 3 neutral
Snowflake’s top-line growth accelerated from 30% to 37% over three quarters.
“over the past three quarters, their top-line growth has gone from 30% to 34% to 37%”
Listen at 3:35
Snowflake’s revenue growth accelerated from 30% to 37% over three quarters.
“over the past three quarters, their top-line growth has gone from 30% to 34% to 37%”
Listen at 3:35
Snowflake’s margins improved as its growth accelerated.
“margins improved while the growth accelerated faster than expected”
Listen at 3:49
Snowflake’s adjusted operating margin rose from 11% to above 15%.
“Adjusted operating margin came in at a little over 15%. It was 11% a year ago.”
Listen at 3:59
Snowflake’s stock has approximately quadrupled since its 2024 low.
“the stock has roughly quadrupled since its 2024 lows”
Listen at 4:18
AI workloads are benefiting Snowflake’s consumption-based business model.
“AI workloads are coming in there and consumption-based model is really, really helping them.”
Listen at 5:43
Snowflake’s AI-driven consumption growth may prove partly temporary and partly sustainable.
“Whether or not it's sustainable or temporary, we'll see, probably somewhere in the middle.”
Listen at 5:50
Snowflake’s valuation is high at about 20 times sales and 80 times free cash flow.
“snowflake right now trades for about 20 times forward sales about 80 times free cash flow that's a lot even with that growth”
Listen at 7:11
Snowflake trades at roughly 20 times forward sales and 80 times free cash flow.
“snowflake right now trades for about 20 times forward sales about 80 times free cash flow that's a lot even with that growth”
Listen at 7:11
Snowflake’s stock-based compensation is nearly 30% of revenue.
“Their stock-based compensation is almost 30% of revenue”
Listen at 7:28
Snowflake’s stock-based compensation is nearly 30% of revenue.
“Their stock-based compensation is almost 30% of revenue.”
Listen at 7:28
Snowflake’s consumption growth could decline if AI usage becomes more efficient.
“Is there a time that maybe the AI volumes get smarter instead of bigger and that consumption-based model kind of comes back down to earth?”
Listen at 8:33
Snowflake’s valuation assumes continued improvement, making it vulnerable to a small setback.
“there's a lot of the status quo continuing and going higher from here baked into that valuation. So any little flinch could cause trouble.”
Listen at 8:44
Snowflake’s high valuation makes it vulnerable to a small deterioration in expectations.
“any little flinch could cause trouble”
Listen at 8:51
Seventy percent of Americans oppose data centers near their homes.
“70% of Americans don't want data centers built near their homes.”
Listen at 13:22
Seventy percent of Americans oppose data centers near their homes.
“70% of Americans don't want data centers built near their homes.”
Listen at 13:22
Eight hundred thirty-three U.S. organizations oppose data center construction.
“there are 833 separate organizations. that are specifically created in the United States to be opposed to data center construction right now.”
Listen at 13:34
Opposition groups blocked or delayed two-thirds of targeted data center projects.
“these 833 groups in the first half of the year, successfully blocked or delayed two out of every three data center projects they targeted.”
Listen at 13:45
Opposition groups blocked or delayed two-thirds of targeted data center projects.
“successfully blocked or delayed two out of every three data center projects they targeted.”
Listen at 13:49
U.S. electricity bills rose about 5%, with larger increases near data centers.
“Electric bills in the United States have risen about 5% on average over the past year, much higher in some areas, specifically the areas near data centers.”
Listen at 14:13
U.S. electric bills rose approximately 5% over the past year.
“Electric bills in the United States have risen about 5% on average over the past year”
Listen at 14:13
Data centers are expected to increase electricity costs another 6% over the next year.
“they're expected to have a further 6% impact over the next year.”
Listen at 14:26
Data centers are expected to add a further 6% impact to electric bills over the next year.
“they're expected to have a further 6% impact over the next year.”
Listen at 14:26
The current data center moratoriums are probably temporary.
“i do think it's probably temporary”
Listen at 15:20
Data center developers can resolve opposition by supplying or contributing power.
“So I do think this problem solves itself by saying, no, supply your own power or even contribute to our grid and bring power.”
Listen at 15:58
Data center disputes can be resolved by requiring developers to supply or fund power.
“I do think this problem solves itself by saying, no, supply your own power or even contribute to our grid and bring power.”
Listen at 15:59
State-level negotiations would end data center moratoriums and restart development.
“I think that that does end these moratoriums and get things going.”
Listen at 16:43
A Louisiana data center project may create 7,500 temporary and 1,000 permanent jobs.
“at peak construction it's estimated to bring 7500 construction jobs which are temporary but this is going to be like an eight-year project as well as about a thousand permanent jobs to the area”
Listen at 17:14
Meta’s Louisiana data center would generate roughly $30 million in property taxes.
“It is bringing in a roughly $30 million property tax bill to a parish in Louisiana whose tax receipts last year were $22 million total.”
Listen at 17:46
A Meta data center would generate roughly $30 million in Louisiana property taxes.
“It is bringing in a roughly $30 million property tax bill to a parish in Louisiana whose tax receipts last year were $22 million total.”
Listen at 17:46
Data center construction will recover.
“I both believe database construction will recover.”
Listen at 19:11
Lumentum is unlikely to increase tenfold again from its current level.
“I don't think we're going to see you know uh lumentum for example 10x again from here because of this.”
Listen at 20:16
AI infrastructure stocks may have passed their biggest gains but have not necessarily peaked.
“I think some of the big gains are behind us when it comes to some of these AI infrastructure companies. But I don't think that we've seen them hit their all-time highs yet, for example.”
Listen at 20:22
Some AI infrastructure companies have not yet reached their all-time highs.
“I don't think that we've seen them hit their all-time highs yet, for example.”
Listen at 20:27
Lou would allocate 60% to an established company and the remainder to a speculative one.
“My answer for me is I'd probably do both. I'd probably say put 60% of the funds in an established company and the rest into a more speculative.”
Listen at 24:52
An investor’s unrealized loss should not affect forward-looking investment decisions.
“The unrealized loss that you have is irrelevant to any forward-looking investment decisions”
Listen at 25:41
Unrealized losses should not affect forward-looking investment decisions.
“The unrealized loss that you have is irrelevant to any forward-looking investment decisions”
Listen at 25:41
A two-to-three-times return by 2030 implies 17% to 29% annualized returns.
“That translates to 17% to 29% annualized returns through 2030.”
Listen at 26:13
A two-to-threefold return by 2030 implies 17%–29% annualized returns.
“That translates to 17% to 29% annualized returns through 2030.”
Listen at 26:13
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.