Good morning. Donald Trump’s car tariffs were felt in the market yesterday, but unevenly. Big US carmakers GM and Ford were down, as were international brands such as Toyota, Honda and Kia. But car rental companies Avis and Hertz had great days, up 20 per cent and 23 per cent, respectively. We would not have guessed that the rental companies would be the ultimate tariff winners. Did you? Shoot us an email if you are smarter than we are: robert.armstrong@ft.com and aiden.reiter@ft.com.
CoreWeave
Later today, if all goes as planned, CoreWeave will complete its IPO and the shares will begin trading. It’s an important IPO, and not just for CoreWeave’s backers. CoreWeave owns data centres built around Nvidia GPUs. Its customers rent capacity in those centres to do AI things. So the IPO is, rightly or wrongly, a live test of the market’s appetite for the AI narrative.
It will test something else, too: investors’ ability to value CoreWeave’s business.
There are some things about CoreWeave that look suboptimal from an investor point of view: it is dependent on Nvidia as primary supplier, investor and sometime customer; heavily dependent on Microsoft as a customer; has a debt-heavy capital structure; there may be an asset-liability mismatch; and it has a slightly odd history. I will not write about these things, because Alphaville’s Bryce Elder covered them all yesterday. Read his pieces and decide for yourself.
But for all this, CoreWeave’s business model is appealingly simple. Again: it builds GPU computing infrastructure and rents it out. Its success or failure depends on the ability, over time, to collect more in rent payments than it costs to finance and operate the infrastructure. It’s a spread business, like a real estate investment trust or a bank.
There are two important things about valuing this sort of spread business, which sound alike but are different: its capital costs, and its cost of capital. Capital costs are investment: what CoreWeave spends on the infrastructure it rents out. This appears on the profit and loss statement on the depreciation line (depreciation is investment spending, spread over the useful life of those investments). Cost of capital is financing cost: what CoreWeave pays for the money it buys the infrastructure with. We will find out CoreWeave’s cost of equity capital tomorrow, when the shares come out. The cost of its debt capital appears on the P&L as interest expense.
So what you don’t want to do is value this sort of company using profit metrics that exclude capital costs and costs of capital, like ebit (earnings before interest and taxes, that is, operating income), or — saints protect you — ebitda (earnings before it interest taxes depreciation and amortisation). You would not value a bank or a Reit with these metrics, because valuing a bank or Reit while excluding cost of debt capital or depreciation expense would be bonkers. You shouldn’t do it with CoreWeave, either.
CoreWeave itself uses ebitda in some contexts; you can search for the term in its prospectus. We’ve also seen brokers throwing around numbers like multiples of ebit. That’s no scandal; it’s standard Wall Street practice. Just don’t do it yourself. What you should do instead is think like someone who’s considering buying an office building. Don’t think about multiples of some form of profit. Does the cash that comes in the door from the tenants cover the operating and financing and capital costs, with a little something left over to put in your pocket?
But with a very fast-growing business like CoreWeave, you can’t look at cash in and out today, because the company is investing heavily to meet rising demand. The company’s free cash flow in 2024 was negative $6bn. So what you have to do is model what future cash rent and cash capital/financing/operating costs will be, with the crucial variable being the demand/supply balance for GPU compute capacity.
This will be hard. While you work on your spreadsheet, we leave you with a question. It’s not rhetorical. CoreWeave’s key customers — the gigantic tech “hyperscalers” — have access to very, very cheap debt capital. CoreWeave has a high cost of debt; some of its debt carries double-digit rates. So why does it make sense, from the point of view of the industry, for CoreWeave to finance the hardware, and then to be reimbursed by its big customers? What does the industry get in return for the extra cost? A good answer to that question will have a lot to say about whether the company can create enduring value.
The fiscal impulse
When US government deficits grow, corporate profits and stock prices tend to grow with them. When the government spends more than it taxes, it forces liquidity into the financial system and money into consumers’ pockets, and from there on to corporate balance sheets. Of course, at some point, out-of-control deficits cause financial crises, and everyone loses. But the US hasn’t hit that point yet.
So, with the federal budget process under way, can equity investors look forward to a positive fiscal impulse — wider deficits — or a negative one?
A few weeks ago Congress passed a budget resolution — a rough outline of the eventual budget. It included plans to keep Trump’s 2017 tax cut on the books and made room for potential new tax cuts (no tax on tips, no taxes on social security benefits), which could cost anywhere from $500bn-$1tn over 10 years. It also contained $200bn-$350bn in new spending to fight illegal immigration. As an offset, lawmakers put in $2tn of proposed spending cuts over 10 years — but we are doubtful they can really cut that much.
If this is what the actual budget looks like, deficits will probably grow — in other words, the fiscal impulse will be positive. And it could be particularly positive in the next 12 months, according to Ed Mills, policy analyst at Raymond James:
Extending out the baseline [tax] policy is kind of the base case; that does not change the fiscal impulse. The total change we will see is probably in the $1tn-$2tn range over the next 10 years, or $100bn-$200bn on a per-year basis. Some of this could be front-loaded, especially [some of the new tax cuts] . . . We may also see some of the spending provisions, like the [Department of Defense] support brought forward . . . Oftentimes, savings or budget cuts are phased in, more on the back end.
The spending will come sooner, the cuts later. So long as this does not cause the bond market to wretch, that should be good for stocks.
But expectations have a role to play. From Marko Papic at BCA Research:
What matters for markets here is the relative change in expectations. If you spent the last five years juicing the economy and market, and then you go from that level of spending to just a mere $100bn-$200bn each year, that will have big implications . . . The dollar and US assets [could] collapse.
When Trump was re-elected, many thought he would reproduce the fiscal profligacy of his first administration. According to the Committee for a Responsible Federal Budget, Trump added $8.4tn to the US deficit — including Covid-19 spending — that Joe Biden built upon in his term, adding $4.3tn (those numbers are contested, including by the Republican-led budget committee). But the crucial point is that the US stock market has benefited from heavy deficit spending over the past few years. An addition of $1tn-$2tn over 10 years amounts to a weaker, if still positive, impulse.
The budget process has to run its course and the bond market has to render its verdict. A lot could change.
But it is worth observing that, on the current trajectory, the US will be trimming its fiscal sails at the same time as Europe and China are letting theirs out, with implication for the dollar and US risk assets. By “rebalancing the global fiscal see-saw”, as Papic puts it, the Trump administration may get what it wants: a cheap dollar, more US exports and a lower-than-expected deficit. If it can bear the market pain.
(Reiter)
One good read
All together now: a value added tax is not a tariff, a non-tariff barrier, or an export subsidy.
Read the full article here