stocksplain

What I'm watching: adoption speed, pricing versus CoreWeave, and whether AWS and Azure answer with their own compute-as-a-service push. Margins in this business tell you how crowded it really is. Could be a real second source. Probably won't be easy.

Selling them externally is the harder problem. CoreWeave got to a $23B valuation by being the GPU middleman when Nvidia chips were scarce. Renting TPUs as a service is a different operational game than running them for your own search engine.

Alphabet ($GOOGL) and Blackstone are spinning up a joint AI compute company, renting out Google's TPU chips to businesses. Direct shot at CoreWeave and Nebius, who built their businesses renting Nvidia GPUs. The AI infrastructure layer is still being carved up. #AI #stocks

@WStreetDragon Yes, I respect the people that buy shares hoping for the permits. I genuinely hope they do get the permits in order which is when I'll probably look at it again

@stocksplain This misses some major points in TMC’s favor but u ain’t wrong

$TMC is a deep-sea mining company. No revenue. No commercial license. 47 employees. The International Seabed Authority has never issued one. Several countries want a moratorium. Funding model: sell shares. Dilution isn't a risk here, it's the business. #DeepSeaMining

Options from here usually rhyme: raise more money (diluting existing shareholders), land a big partner, or run out of clock. A business that feeds itself vs. one that needs someone else to keep the lights on. $TMC is firmly the second one.

Burning cash isn't automatically fatal. Amazon did it for years. The question is runway: enough cash to survive until revenue shows up? $TMC's ~$22M doesn't buy much time at current burn.

Free cash flow = cash left after running and building the business. $TMC, a deep-sea mining hopeful, has roughly $22M cash, ~47 employees, and basically zero revenue. Classic negative FCF: every dollar out, almost nothing in. #InvestingBasics

Debt-to-equity ratio: how much a company owes vs. what it owns. $S (SentinelOne, a cybersecurity firm) sits near zero. Sounds healthy. Mostly it just means they're funded by investors, not lenders. The real risk isn't debt, it's whether revenue ever catches spending. #investing

The quieter watch: any sign hyperscaler AI spending is digesting. The pace normalizes eventually. May 20 is one data point. At $5.4T market cap, the valuation assumes a lot goes right. China returning would just be one more thing.

Nvidia has a pattern of beating estimates that already sit above its own guidance. So 'beat' alone tells you nothing. Margin trajectory and data center commentary carry more signal than the headline number.

$NVDA reports FQ1'27 earnings May 20. Consensus: 80% revenue growth. Strip out China, where US export rules have locked them out, and it's 105%. The real question isn't the beat. It's what they say about China access going forward. After close, May 20. #NVDA #Earnings

One print tells you little. $IBM spun off Kyndryl in 2021 and has been reshaping itself around software and AI tooling for years. EPS trends across several quarters say more than any single quarter ever will.

Two versions matter: basic EPS and diluted EPS. Diluted accounts for stock options and convertible debt, anything that could quietly create more shares. It's the more honest number. Always check which one you're reading.

EPS = profit ÷ shares. $IBM (software, consulting, mainframes for banks and hospitals) posted ~$3.92 last quarter. That's your slice of earnings per share, not cash you receive. At ~23x earnings, the market's paying $23 for every $1 of profit.

The further a company is from actual profitability, the more its forward P/E depends on imagination rather than evidence. Not inherently wrong. Just a different kind of risk than most P/E ratios quietly carry.

$IONQ is a quantum computing company that lost money in 2024. So what does its forward P/E even mean? Trailing P/E uses real past earnings. Forward P/E uses analyst guesses for next year. When earnings are negative, only the guess works, and the guess is doing a lot of work.

Trailing P/E uses the last 12 months of actual earnings. Forward P/E uses analyst estimates for the next 12. $IONQ builds quantum computers and has no profits yet, so trailing P/E is undefined. Forward P/E exists, but it's a guess on a guess. #investing101