In partnership with
Reports of semiconductor companies and electric power equipment companies being sold out till at least 2028 due to huge AI data center orders are pretty common these days. Stocks of these companies have rerated and have gone parabolic, creating enormous wealth for investors who spotted this trend early. Kudos to them; they deserve it. However, the question now is: are investors overvaluing the companies?
I have said this before, and I will say it again: the AI revolution is here to stay. However, investors always push asset prices beyond their true economic value during the initial gold rush. This is reflexivity. That is why we have periods of overvaluation and undervaluation.
Note
The theory of reflexivity in financial markets — popularized by George Soros — is that investor biases and perceptions can influence economic fundamentals, creating self-reinforcing feedback loops that drive prices away from equilibrium. Essentially, market participants' distorted views can actively shape market outcomes.
I shared my thoughts about the CAPEX mania earlier this month when I covered a few companies, specifically Lumentum. In short, innovation in AI algorithms would eventually drive down compute requirements and costs. Also, companies like OpenAI have no moat, can be disrupted within a few months by a competitor with a better model, and there are no switching costs. These facts bring into question whether these commodity companies can turn profitable and justify their CAPEX.
The whole AI ecosystem — especially the companies that have gone parabolic — is one press release away from coming back down to earth. If one of the bigger players pulls back on CAPEX, AI-hyped stocks come back down to reality.
While one can be bullish on AI over the long term, it would be prudent to invest in AI-related companies at realistic valuations. One of the AI-related themes that I have been extremely bullish on is the electric power space; companies that enable and support the electricity grid.
Companies like GE Vernova (GEV), which has the largest installed base of gas turbines, steam turbines, and generators of any OEM in the world, are sold out till 2028. Their order backlogs have been soaring. For instance, GEV’s order backlog has grown from $116 billion in 2024 to $163 billion in Q1 2026 and is expected to reach $200 billion in 2027. With improving operations and a rapidly growing backlog, the stock has done well since it started trading as a separate entity in April 2024. GEV is up 865% since then.
Being sold out for years on out and having a huge backlog is awesome. For investors, what matters most is the revenue conversion of that backlog and pricing power that leads to higher margins. So, investors need to look at the following:
What is the book-to-bill ratio? This is essentially new orders received divided by orders shipped and billed. A high book-to-bill ratio indicates solid demand but may require higher CAPEX and G&A (more staff) to meet the demand. If a company is not prudent with its outlays, it can end up overbuilding and overhiring to meet demand, which may be temporary or artificial if customers wrongly forecast their own needs.
What is the Backlog Ratio? This is calculated by dividing the Total Backlog by Quarterly Sales. In cases where a company is completely sold out, the backlog ratio is how many quarters the company would need to fulfill the orders.
Pricing of new orders or slot reservation agreements. Higher pricing would indicate pricing power and potential ability to generate higher margins.
The whole exercise is not to get caught up in the hype and avoid making the mistake of chasing an overvalued asset. With that said, let’s take a quick look at GE Vernova.
GE Vernova
GE Vernova (GEV) is a global leader in the electric power industry, with products and services that generate, transfer, orchestrate, convert, and store electricity. GEV reports financial results across three business segments:
Power - includes the design, manufacture, and servicing of gas, nuclear, and hydro technologies, providing a critical foundation of dispatchable, flexible, stable, and reliable power.
Electrification - includes power transmission, grid systems integration, power conversion and storage, and grid automation and software technologies required for the transmission, distribution, conversion, storage, and orchestration of electricity from the point of generation to the point of consumption.
Wind - includes our wind generation technologies, both onshore and offshore wind turbines and blades.
GEV Tailwinds
The AI electricity supercycle is structural, not cyclical. Even with reported data center buildout delays in 2026, the global data center electricity consumption is expected to double.
Due to higher electricity demand for data and computing needs, and governments for national security and economic growth, the global electricity consumption is expected to double from 95 exajoules in 2024 to 170 exajoules in 2050. Hence, GEV is not just an AI beneficiary.
On April 26, under section 303 of the Defense Production Act of 1950, President Trump announced that electricity grid infrastructure is essential to national defense. This covers transformers, transmission lines and conductors, substations, high-voltage circuit breakers, power control electronics, protective relay systems, capacitor banks, electrical core steel, and related raw materials and manufacturing tools. While full details are awaited, this will positively impact GEV.
GEV completed Prolec GE’s acquisition on February 2, 2026. Prolec makes GEV the only company in the world capable of supplying an integrated ‘power-to-rack’ solution. Power-to-rack covers the gas turbine, the distribution transformer, and the switchgear. This vertical integration capability is in high demand from hyperscaler customers.
Sold out till 2028. Has a $163 billion backlog growing to $200 billion in 2027.
GEV’s new orders and slot reservation agreements in 2026 are priced 10 to 20 points above the backlog pricing entered into in 2025.
Quarter to date, we have booked more power equipment orders in terms of value than we did in all of Q1 2026. On pricing, we expect our orders in the first half of 2026 to be priced 10-20 points higher than our 4Q 2025 orders on $1 per kW basis.
Overall, we see industry demand for grid resiliency products as growing low double digits through the end of the decade.
Quite the list of tailwinds for GEV; however, is GEV overvalued at 40× 2026 EV/EBITDA? One could argue that, given the supercycle nature of tailwinds behind GEV, one should assign a higher multiple than has historically been assigned (15x-to-18x NTM) to a high-quality cyclical OEM like GEV. If one were to use an 18x EV/EBITDA multiple, GEV is overvalued by 55%.
Irrespective of that, let’s look at other factors before jumping to any conclusions.
Net Income Mirage
GEV Quarterly Income Statement Highlights
GEV’s revenue growth was disappointing for Q4 2025 (quarter ended Dec 2025); however, the market overcame that disappointment as the focus is on GEV’s exploding order backlog well into 2030 and beyond, and YoY revenue growth in Q1 2026 didn’t disappoint.
What was surprising was that the market ignored the deterioration in gross margin and operating margin.
A good chunk of GEV’s Income before Taxes comes from non-operating income or other income. This income is mainly from financial interests in China XD Electric, Aero Alliance, Hitachi-GE Nuclear Energy, and Prolec GE (now fully part of GEV). There is nothing untoward here, having diverse 20+% interest in joint ventures. Investments made by GEV’s financial services business also contributed to other income.
The huge bump in Net Income in Q4 2025 stands out. The main contributor was an income tax benefit. Hence, this was a one-off. Here is the note from the 2025 10K; make what you will of it.
We recorded an income tax benefit on pre-tax income for the year ended December 31, 2025, primarily due to a decrease in valuation
allowances from a change in judgment regarding the realizability of a significant portion of our U.S. federal and state deferred tax assets.
Source: Page 28, FY2025 10K
Another thing that stands out is the huge bump in other income in Q1 2026. Non-recurring items caused the increase in other income from the trend, as most of it relates to Prolec GE. Here is the note from the Q1 2026 10Q.
Includes a pre-tax gain of $3,992 million related to the acquisition of the remaining 50% stake in Prolec GE from Xignux as a result of the remeasurement of our previously held equity interest to fair value, ...
The point here is that if an investor were to use the earnings from the last two quarters as a trend, they would draw the wrong conclusions about GEV. For instance, GEV’s trailing P/E is currently 33. However, if one were to back out the one-time gains, the trailing P/E would be ~120. That is quite a P/E for a business projected to grow low double digits with abysmal operating margins.
Poor FCF Growth
While the one-time items boosted Net Income, the good thing is that they are subtracted from Net Income while calculating Cash from Operations. Cash from Operations (CFO) looks stellar for both Q4 2025 ($2.5 Billion) and Q1 2026 ($5.2 Billion) when compared to the CFO of $967 million in Q3 2025 and $367 million in Q2 2025. The stellar numbers in Q4 2025 and Q1 2026 were driven by new collections received in excess of revenue recognition.
Now, here is the catch. The midpoint of GEV’s full year 2026 Free Cash Flow (FCF) guidance is $7.0 billion. Driven by stellar CFO of $5.2 billion in Q1, GEV’s FCF came in at $4.8 billion. That would imply total FCF for the next three quarters to be $2.2 billion (7.0 - 4.8) or $734 million per quarter on average. Quite the come down with no major CAPEX announced.
It gets more interesting when one takes into account GEV’s cumulative FCF guidance for the three years covering FY2026 to FY2028. In its Q4 2025 presentation, GEV guided for $22 billion in cumulative FCF for those three years. That is $7.34 billion on average per year. Clearly implying that FCF will not improve materially till FY2028 and anything beyond is, well, just a plan.
So, what have we till now? An extremely high P/E once one backs out one-time gains and flat FCF for the next three financial years. So, is 40× 2026 EV/EBITDA justified? Let’s hold that thought and address the backlog hype.
Backlog Conversion
Backlog is reported as Remaining Performance Obligation (RPO). Consider the following note from the Q1 2026 10Q.
As of March 31, 2026, the aggregate amount of the contracted revenues allocated to our unsatisfied (or partially unsatisfied) performance obligations was $163,276 million. We expect to recognize revenue as we satisfy our remaining performance obligations as follows:
(1) Equipment-related RPO of $75,924 million, of which 38%, 69%, and 98% is expected to be recognized within 1, 2, and 5 years, respectively, and the remaining thereafter.
(2) Services-related RPO of $87,352 million, of which 16%, 53%, 78%, and 91% is expected to be recognized within 1, 5, 10, and 15 years, respectively, and the remaining thereafter.
Remember, GEV is sold out till 2028 and is now only taking orders for 2029 and 2030. So, RPO conversion has to align with guidance. Let’s do the calculation for FY2026:
Equipment-related RPO conversion = 75,924 × 0.38 = 28,851
Services-related RPO conversion = 87,352 × 0.16 = 13,976
Total RPO conversion FY2026 = 28,851 + 13,976 = $42,827 million or ~$43 billion
FY2026 Revenue Guidance mid-point = $45 billion
Close enough. Now, where did that extra $2 billion in guidance come from? Possibly from expectations of more service-related revenue and some boost from Prolec GE as it is now fully part of GEV; I don’t know, but there is a possibility of a revenue miss if anything slightly goes haywire.
Now let’s do FY2027:
Equipment-related RPO conversion = 75,924 × (0.69 - 0.38) = 23,536
Services-related RPO conversion = 87,352 × (0.53 - 0.16) / 4 = 8,080
As services-related RPO is projected for 1 and then 5 years, I took the average of four years to arrive at the FY2027 conversion number and liberally didn’t backload it.
Total RPO conversion for FY2027 = 23,536 + 8,080 = $31,616 or ~$32 billion
That is $11 billion less than FY2026 as calculated above. Again, since GEV is sold out till 2028 for critical power-generation equipment, specifically gas turbines, power transformers, and switchgear, GEV’s Electrification and Wind segments will have to do the heavy lifting to help GEV’s total revenue grow in low double digits in FY2027. Possible, but the wind segment has been an underperformer (regulatory hurdles, projects delayed, political headwinds, etc.) and is expected to continue to drag. Hence, the heavy lifting will squarely fall on the electrification segment. Investors betting on a boost from AI demand, which is not coming, could be setting up for a big shock in FY2027.
Verdict
So, we have GE trading at an extremely high P/E ~120 when one backs out one-time gains, flat FCF till the end of FY2028, and possible revenue underperformance in FY2027, along with no further AI-related demand boost as GEV is sold out.
Again, is 40× 2026 EV/EBITDA justified for GEV? I don’t think so. Given the risks, even after considering GEV’s structural long-term growth potential due to overall demand in the power and electrification segments well into the next decade and its leadership position, I would be willing to assign a 15x 2026 EV/EBITDA. That would imply a fair value of 410. At 18x it would be 492. Current share price is 1093.
I would avoid investing in GEV at this time as I consider it extremely overvalued. I would consider adding GEV to my portfolio if it falls to 410. This is just my opinion; I leave you to do as you desire based on the facts as presented.
Quick Updates
Last year, I wrote that BJ (BJ's Wholesale Club), DPZ (Domino's Pizza), and TSCO (Tractor Supply Company) would underperform well into this year. All of them are down ~30% since I wrote about them. DPZ and TSCO both reported earnings recently and missed estimates, resulting in the stocks falling 10% after reporting.
I request that new subscribers read the articles from May to August of last year, as a lot of what I wrote then is still in play.
I wrote positively about FDX (FedEx) recently, and it is nicely up ~10% since then.
As I was writing this post, I saw the news about OpenAI missing its own targets for new users and revenue; stumbles that have raised concern among some company leaders, especially the CFO, about whether it will be able to support its spending on data centers. And then I saw the market selling of semiconductor names. My first reaction was: ROI has entered the chat.
Until next time. Have fun!!!
ChatGPT gives you generic answers because you give it generic prompts.
You know the fix: longer prompts, more context, clearer constraints. But typing all that takes five minutes per prompt, so you shortcut it. Every time.
Wispr Flow lets you speak your prompts instead of typing them. Talk through your thinking naturally — include context, constraints, examples — and get clean text ready to paste. No filler words. No cleanup.
Works inside ChatGPT, Claude, Cursor, Windsurf, and every other AI tool. System-level, so there's nothing to install per app. Tap and talk.
Millions of users worldwide. Teams at OpenAI, Vercel, and Clay use Flow daily. Free on Mac, Windows, and iPhone.
